research paper on mobile database

The Research and Application of Embedded Mobile Database

Ieee account.

Change Username/Password
Update Address

Purchase Details

Payment Options
Order History
View Purchased Documents

Profile Information

Communications Preferences
Profession and Education
Technical Interests
US & Canada: +1 800 678 4333
Worldwide: +1 732 981 0060
Contact & Support
About IEEE Xplore
Accessibility
Terms of Use
Nondiscrimination Policy
Privacy & Opting Out of Cookies

A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity. © Copyright 2024 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.

Academia.edu no longer supports Internet Explorer.

To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to upgrade your browser .

Enter the email address you signed up with and we'll email you a reset link.

We're Hiring!
Help Center

Mobile Database System: Role of Mobility on the Query Processing

2010, Computing Research Repository

The rapidly expanding technology of mobile communication will give mobile users capability of accessing information from anywhere and any time. The wireless technology has made it possible to achieve continuous connectivity in mobile environment. When the query is specified as continuous, the requesting mobile user can obtain continuously changing result. In order to provide accurate and timely outcome to requesting mobile user, the locations of moving object has to be closely monitored. The objective of paper is to discuss the problem related to the role of personal and terminal mobility and query processing in the mobile environment.

Related Papers

Mobile Information Systems

nisar Ahmed

Information Sciences

H. Grine , S. Lecomte

Pascal Molli

Abstract This paper reports on the main results of a specific action on mobile databases conducted by CNRS in France from October 2001 to December 2002. The objective of this action was to review the state of progress in mobile databases and identify major research directions for the French database community.

Operations Research/Computer Science Interfaces Series

Sanjay Madria

Kshama Raichura

Rohiza Ahmad

Alexander Decker

James Jayaputera

EDUARDO AMABLE SAMANIEGO MENA

Mobile learning: research context, methodologies and future works towards middle-aged adults – a systematic literature review

Track 5: Multimedia and Education
Published: 20 August 2022

Cite this article

Syahida Mohtar ORCID: orcid.org/0000-0002-4462-8890 1 ,
Nazean Jomhari 1 ,
Mumtaz Begum Mustafa 1 &
Zulkifli Mohd Yusoff 2

Over the past several years, mobile learning concepts have changed the way people perceived on mobile devices and technology in the learning environment. In earlier days, mobile devices were used mainly for communication purposes. Later, with many new advanced features of mobile devices, they have opened the opportunity for individuals to use them as mediated technology in learning. The traditional way of teaching and learning has shifted into a new learning dimension, where an individual can execute learning and teaching everywhere and anytime. Mobile learning has encouraged lifelong learning, in which everyone can have the opportunity to use mobile learning applications to gain knowledge. However, many of the previous studies on mobile learning have focused on the young and older adults, and less intention on middle-aged adults. In this research, it is targeted for the middle-aged adults which are described as those who are between the ages of 40 to 60. Middle-aged adults typically lead very active lives while at the same time are also very engaged in self-development programs aimed at enhancing their spiritual, emotional, and physical well-being. In this paper, we investigate the methodology used by researchers based on the research context namely, acceptance, adoption, effectiveness, impact, intention of use, readiness, and usability of mobile learning. The research context was coded to the identified methodologies found in the literature. This will help one to understand how mobile learning can be effectively implemented for middle-aged adults in future work. A systematic review was performed using EBSCO Discovery Service, Science Direct, Google Scholar, Scopus, IEEE and ACM databases to identify articles related to mobile learning adoption. A total of 65 journal articles were selected from the years 2016 to 2021 based on Kitchenham systematic review methodology. The result shows there is a need to strengthen research in the field of mobile learning with middle-aged adults.

Avoid common mistakes on your manuscript.

1 Introduction

Adulthood can be categorized into early, middle and late adulthood. Middle-aged adults come between the ages of 40 to 60, in other words is when one is in between the younger and older generations [ 42 , 62 ]. This stage of age, notably is the age period that Hall [ 31 ] referred to as aging, where the signs of cognitive and physical ageing start to be noticeable, from the age of 40 and rapidly increase after the age of 65 [ 6 , 54 ]. According to Yaffe and Stewart [ 94 ], a large part of adult life is made up of the mid-life period. This has been associated with many descriptive terms: mid-life syndrome, mid-life crisis, middlescence, empty nest syndrome, second adolescence, second honeymoon, age of fulfillment, and menopause. Aging population contributes to healthcare issues, not only amongst the older adults but towards middle-aged adults too. As mentioned earlier, the healthcare issues amongst the middle-aged adults are related to the decline in physical abilities, relational, and psychological capacities. For example, women in their middle age experience menopause and perceived personality change, which lead to severe depression, physical, and emotional problems [ 80 ]. According to Yaffe and Stewart [ 94 ], the most frequently identified events or concerns among middle-aged adults were: increased personal concern for health, death of a friend or relative, change in wage/salary, and concern for change in physical appearance.

When middle-aged adults enter their 60s, their reaction time starts to slow down further, and they experience a significant declination in their performance. The brain may also no longer function at its optimal level, leading to problems like memory loss, dementia, and may have issues with other cognitive functions such as language, attention, and visuospatial abilities [ 35 , 61 ]. It has been widely assumed that the midlife period is a critical period, thinking about death and mortality, as well as experiencing decline in physical abilities, relational, and psychological capacities [ 80 ]. Therefore, early prevention should therefore be looked upon at the middle age stage to help with memory impairment, as well as emotional control.

Middle-aged adults typically lead very active lives while also engaging in self-development programs aimed at enhancing their spiritual, emotional, and physical well-being. Muslim adult, for instance, will prefer to go to the mosque, surau, or Islamic center to seek for Islamic education [ 42 ] to enrich their knowledge and gain serenity through the command of Islam. This indicates that an individual in the middle-aged is inclined to reflect and improve the quality of one’s daily practice. Unfortunately, during the Covid-19 pandemic outbreak, many lectures at the mosques and other institutions could not be held, resulting in many people having to work from the home. As a result, many have taken the initiative to hold religious lectures online through video conferences such as via Zoom, WebEx, Jitsi Meet, Google Meet applications, and many more [ 1 ]. There are also those who watch religious lectures that have been prerecorded on certain channels, such as YouTube or podcasts. However, the enthusiasm and motivation for online and prerecorded learning is not the same and less encouraging as compared to face-to-face lectures.

Health management apps have shown to be useful for treating a variety of illnesses such as chronic illnesses caused by obesity, high blood pressure, diabetes, and so on [ 32 ]. As middle-aged adults are smartphone and tablet active users, they can use these portable devices to track their healthy lifestyle habits, maintain social communication, prevent accidents, and seek information [ 91 ]. In addition to chronic illness management using mobile applications, there is also a concern on how middle-aged adults can utilize mobile technology in fulfilling their spiritual journey towards a quality lifestyle. For example, they can learn how to acquire a literal understanding of the Quran through a spiritual mobile application. This will help a Muslim to elevate their understanding, motivation, and devotion towards Islam, which eventually leads them to become a better person emotionally and psychologically. All of these exhibit many important experiences associated with middle-age adults, most involving work and family, and self-development [ 53 ].

Mobile devices such as smartphones have gained popularity because they allow people to stay in touch and provide easy access to information anywhere and anytime [ 89 ]. Therefore, investigating the acceptance and adoption of mobile learning by the middle-aged adults through a systematic literature is important in highlighting the gap for any future work.

This review paper presents the fundamentals of mobile learning and the utilization of mobile technology in the learning environments. Mobile learning theories are also highlighted to show the significance of mobile learning towards middle-aged adults. Based on the research context found in the selected literature, the researchers here provide a systematic mapping of the employed methodologies in the area of mobile learning research. The purpose of the systematic mapping is to determine the most appropriate methodology for future research on middle-aged adults in areas of mobile learning.

2 Mobile learning

M-learning is a subset of ‘e-learning’ while ‘e-learning’ is the subset of distance learning that focuses on learning across context and learning with mobile devices, which can take place anytime, anywhere [ 43 , 62 ]. For example, learning may happen at the workplace, at home, and at places of leisure. The learning may be related to work demands, self-improvement, or leisure; and it is mobile with respect to time where it happens at different times during the day, on working days, or on weekends [ 68 ].

According to Ozdamli and Cavus [ 70 ], learners, teacher, environment, content, and assessment are the basic elements of mobile learning. The core characteristics of mobile learning are ubiquitous, portable size of mobile tools, blended, private, interactive, collaborative, and instant information. They enable learners to be in the right place at the right time, that is, to be where they can experience the authentic joy of learning.

Since learning can be performed anywhere and anytime using electronic devices, Traxler [ 85 ] defines that mobile learning is a learning process that is delivered through the support of mobile devices such as personal digital assistants, smartphones, wireless laptops, and tablets. This understanding is supported by Keegan [ 45 ] who suggested that m-learning should be restricted to learning on small and portable devices as mobile devices that could be carried everywhere.

According to Nordin, et al. [ 69 ], the requirements for mobile learning environment include technology, that is, (1) highly portable (to support learning whenever and wherever), (2) individual(the design should be able to support individual learning, cater for individual learning styles and be adaptable to learners’ abilities), (3) unobtrusive(where learners should be able to retrieve knowledge without the technology becoming a deterrent), (4) available(enabling communication with friends, experts and/or teachers), (5) adaptable(the context of learning should be adaptable to situations and the individual’s skills and knowledge development), (6) persistent( able to manage the learner’s learning despite the changes in the technology itself), (7) useful(useful to learners for everyday chores), and (8) user-friendly(easy for people to use and must not create technophobia among new users).

3 Mobile technology

Today, it is fortunate that mobile technology’s on-demand capability puts learning back into the learner’s hands by allowing users to take the initiative in diagnosing their learning needs, formulating learning goals, identifying human, and material resources for learning, choosing and implementing appropriate learning strategies, and evaluating those learning outcomes [ 50 ].

Mobile technology covers a wide range of mobile devices such as portable electronic devices used to perform a wide variety of communication, business, productivity, and lifestyle tasks such as parenting [ 26 , 66 ]. It is also connected through a cellular communication network or a wireless connection. The common mobile technologies that allow these tasks are cellular phones, PDAs, handheld computers, tablets, laptops, and wearable devices. A standard mobile technology device, such as a cellular phone, may have one or more features such as a GPS, a web browser, an instant messenger system, an audio recorder, an audio player, a video recorder, and gaming systems [ 4 ].

In the area of healthcare, numerous studies have been conducted on the use of mobile devices with wearable devices [ 21 , 39 , 87 ] to monitor the health of the elderly and individuals with disabilities. By using mobile apps, the health of the elderly and young adults can also be tracked [ 12 , 20 , 40 ] and diagnosed using mobile game-based screening tools [ 34 ], especially when facing challenges and stressful time during the Covid-19 pandemic [ 79 ] , or during post-college life transition [ 27 ]. Not all older people are proficient in using mobile devices. Therefore, there are researchers who make studies related to how older and young adults (university students) manages their mobile device security and privacy settings of their mobile devices in the context of social interaction and motivation [ 64 , 67 , 90 ].

Besides the usage of mobile devices in the healthcare area, the growth of mobile devices is significant and impactful in the education area such as in teaching history using 3D [ 57 ] and safety education [ 13 ], personal learning and workplace learning [ 29 ]. The use of mobile devices such as smartphones and tablets has become truly ubiquitous and has a potential for improving student learning, which can happen in collaborative, authentic settings, i.e., real life contexts and use active learning approaches [ 18 ]. As smartphones have become popular devices among youth nowadays [ 36 , 65 ], these devices can be utilized and embraced in the classroom teaching environment. By having a smartphone with wi-fi connectivity, Bluetooth, camera, color display, audio/video recording capability, it is already suitable for a person to adopt m-learning [ 36 ]. Majority of students spend most of their time (6 to 24 hours) on the Internet using their smartphones [ 8 ].

Smartphones also have become essential communication tools for older adults to stay connected with their family and peers [ 93 ]. Compared with younger adults, older adults tend to be more likely to use mobile phones for their original design purpose—that is, making calls for instrumental reasons such as arranging plans and other instrumental activities rather than playing games, surfing the internet, or using auxiliary applications [ 91 ]. The intervention of mobile technology in older adults’ lifestyles can improve their well-being and keep their mind and body active as well as prevent or slow down cognitive decline. For instance, mobile games can be used to capture cognitive learning outcomes and the process of knowledge acquisition [ 92 ]. Through activities such as interacting with easy games [ 71 ], taking and managing photographs, sending messages via SMS, video or audio calls, and reading newspapers via webpages may help cognitive and noncognitive stimulation of older adults.

Mobile computing devices become more situated, personal, collaborative, and lifelong and these innovations will become embedded, ubiquitous, and equipped with enhanced features for rich social interaction, contextual awareness, and access to the Internet. Hence, extending learning outside the classroom and into the learner's environment, mobile learning can have a significant impact on middle-aged adults. However, based on the research context in the areas of mobile learning, existing studies have concentrated exclusively on aspects of the mobile device use, such as accessibility, usability, and adoption, among young and older adults, while middle-aged adults have received less attention. Thus, the use of mobile devices among middle-aged adults should be further investigated to determine how mobile devices can assist them in acquiring knowledge and developing themselves while leading hectic lifestyles and having to deal with the Covid-19 pandemic, towards their long-life wellbeing.

4 Multimedia in Mobile learning application

Using mobile device as a learning tool is a new way for learners to learn as they like, anywhere and anytime. Moreover, an application that contains multimedia elements such as text, animation, graphic and video will engage and attract the attention of the student. Mobile learning application used in mobile learning environments varies, such as Learning Management System (LMS), Short Messaging services (SMS), Podcasting, Social Networking, Instant Messaging, Blogging, Facebook, Microblogging, Wiki, QR, 3D and Augmented Reality [ 81 ].

SMS and videos have long been used as language learning tool through the use of mobile phones and personal digital assistants (PDAs) [ 68 ], and today, many have benefits from using WhatsApp, flashcards and mind maps, on-line videos, and social networks in learning. Recently, Duolingo is said to be a popular application for new language learning where learners can interact with intelligent chatbots that give corrective feedback and awards at the same time [ 49 ].

In the fast-aging population countries like China, senior users have become a significant new growth point that cannot be ignored in social network sites to keep continuous competitiveness. In China, WeChat is the most popular social software for senior citizens. This is due to the good user experience and operability, where some senior users manage to operate the application although they have no computer skills or they know little about the network [ 11 ].

On the other hand, instant messaging apps such as WhatsApp and Line have become a popular mobile app amongst students. In a classroom environment, the student may use these apps to interact with teachers outside the class and using smartphones to manage their group assignments. The use of instant messaging applications promotes collaborative learning [ 7 ] and flexible learning, improves student participation, increase communication and interaction between lecturers and students, as well as improve the performance of teaching and learning [ 10 ].

Text editors such as the Mobile MS office, content management systems such as Learning Management Systems (LMSs), and audio-video recording of lectures did not get much attention by the students in terms of its usage via the smartphone. The reason for the low usage of these functions and features could be due to the limited screens space, which makes it difficult to read large documents, and the small sized keypad makes data entry cumbersome [ 36 ]. To make mobile learning more interesting, game-based elements have been used to improve the students’ engagement and enjoyment in learning. For instance, Kahoot is a game-based technological platform that can be accessed from, for instance, smart devices or a laptop. The game-based learning application (app) can benefit working adults who are adult learners with diverse learning abilities. Chunking method was used to break down complex concepts into smaller parts in the form of multiple-choice questions. The students’ learning process is tested and corrected, in real time, through the statistics which are generated from this chunking process. Kahoot creates a safe environment for students to make mistakes through multiple choice questions, and yet relearn it without being judged by their peers. However, the drawback of Kahoot is, it does not adequately support the learning experience of adult learners [ 74 ].

To achieve a successful ageing life, positive spirituality indeed has a close relationship with physical and mental abilities. There have been studies that develop an Empathic-Virtual Coach (VC) to involve senior users in enjoying a healthy lifestyle with respect to diet, physical activity, and social interactions, while in turn supporting their carers [ 41 ]. Furthermore, in addition to physical support, adults also require emotional and spiritual help for a balanced lifestyle. For example, Sevkli, et al. [ 75 ] in their study had designed and developed mobile Hadith Learning Systems (HLS) that were able to encourage and promote hadith learning for young and middle-aged Muslims. Hence, mobile apps appear to be one of the tools that can be used to promote a balanced well-being lifestyle for the older people such as their social status, independence in their everyday activities, health status, standard of living, or leisure activities of the aging population.

5 Mobile learning theory

According to Lee, et al. [ 56 ], there is an increasing number of adult learners entering or returning to university. Despite the growing number of nontraditional adult students in online higher education, little is known about the dynamic processes of adult distance learning, through which adult students struggle to develop their learning ability to balance their life and study, and to become self-regulated learners, and ultimately as competent selves and lifelong learners. The implementations of mobile learning are supported and guided by theories such as Behaviorism, Cognitivism, Constructivism, Situated Learning, Problem-Based Learning, Context Awareness Learning, Socio-Cultural Theory, Collaborative Learning, Conversational Learning, Lifelong Learning, Informal Learning as well as Activity Theory, Connectivism, Navigationism, Location-based learning [ 46 , 68 ]. The classification of activities around the main theories and areas of learning relevant to learning with mobile technologies are shown in Table 1 .

Lifelong learning happens not only in learning institutions such as community colleges or higher learning institutions, but can also happen anytime and anywhere according to the needs of the individual [ 69 ]. Informal and lifelong learning are often referred to adult education or continuing education, which means a learning process that occurs as blended learning with everyday life unobtrusively and seamlessly [ 73 ]. The unique characteristic of lifelong learning is the fact that it is centered on the learner. Because of that, the use of technology in offering a flexible learning framework is often favored by adult learners [ 69 ]. In addition, when compared to conventional methods such as textbooks, mobile learning tools, especially learning through mobile apps, are intrinsically inspiring, provide greater satisfaction, increase student well-being, and have positive implications for long-term student persistence [ 78 ].

Lifelong adult learners are different from young learners (school or university students) who may devote significant amounts of time to study each day, as their learning time is scattered due to family responsibilities, work obligations, and other social obligations [ 44 ]. However, the keys to unlocking the secrets to successful adult learning online are embedded in the basic principles that guide adult learners. The subsequent six principles upon which Knowles [ 51 ] constructed his formal and andragogical concept are shown in Table 2 .

6 Methodology

This study carried out an extensive literature review to identify the research gap, focusing on the related literature published within the period of 2016 to 2021. The aim of this systematic review is to investigate the trend of previous research on the acceptance and adoption of mobile learning by middle-aged adults. In order to justify the research gap based on the previous studies, this article will also provide views on the existing mobile learning usage targeted at solving user’s adoption of mobile learning towards young and older adults.

To conduct the systematic review, the researchers followed the procedure defined by Kitchenham [ 48 ], which is one of the most complete and suitable methods for reviewing studies in computer science. We carried out this review in three main phases: 1) planning of systematic mapping; 2) conducting the review; and 3) reporting the review. The phases of this systematic review and the related activities are shown in Fig. 1 .

Phases of conducting this systematic review

Planning of the Systematic Mapping

Activities involved in this stage were aimed to identify the objectives of the review. These activities are as follows:

Discovering the gap of the existing systematic reviews

In this step, a comprehensive search was performed in the cyberspace to locate the related review studies in mobile learning. Some of the bibliographic databases accessed included EBSCO Discovery Service, Science Direct, Google Scholar, Scopus, and IEEE.

Specifying the research questions

The research questions we have formulated for this review attempt to acquire the understanding and to determine the research gap on mobile learning usage in assisting lifelong learning in the context of spiritual among middle-aged adults. These questions are related to the acceptance and adoption of mobile learning towards middle-aged adults. The research questions are:

What are the fundamentals and background of mobile learning in the learning environment, including its adoption, acceptance, and available applications?

What are the research methodologies employed in the current studies carried out in mobile learning field?

What are the core research gaps should be further investigated by researchers in mobile learning towards middle-aged adults?

Identifying the relevant bibliographic databases

To answer the research questions and find relevant studies, bibliographic databases that cover majority of journals and conference papers associated with the field of human-computer interaction and mobile learning were selected. Related literatures published within the period of 2016 to 2021 were chosen in this research and the relevant bibliographic databases are ACM ( https://www.acm.org/ ), Emerald ( https://www.emerald.com/insight/ ), EBSCO Discovery Service ( http://search.ebscohost.com ), Science Direct ( http://sciencedirect.com ), Google Scholar ( http://scholar.google.com ), Scopus ( http://scopus.com ), and IEEE ( http://ieee.com ).

Conducting the Review

Activities involved in this stage were aimed to selecting related studies. These activities are as follows:

Identifying the Relevant Studies

In identifying the relevant studies, a search using key words such as “human-computer interaction”, “mobile learning”, “middle-aged adults”, “us- ability” was conducted. Accordingly, Boolean OR was used for alternative spellings, synonyms, or alternative terms, and Boolean AND was applied to connect the main terms. The complete list of search keywords of the review is provided in Table 3 .

Two additional search strategies were applied to retrieve the maximum number of relevant papers. The first strategy was reviewing the reference list of selected papers to find more related papers. The second strategy was googling the authors of selected studies to find potential related research.

Defining Selection Criteria

For selecting the primary papers, the following criteria based on the purpose of this study are defined.

Inclusion Criteria:

Studies containing mobile learning, acceptance, and adoption among mobile devices users.

Studies dealing with factors that contribute to the adoption and acceptance of mobile learning in the educational environments or working environments.

Studies utilizing mobile learning applications related to education, health care, data collection, and engineering that motivate users to use mobile learning.

Studied involving mobile learning users in category young adults, middle-aged adults, and older adults.

Exclusion Criteria:

Studies in learning environments that do not relate to the mobile learning context.

Studies of mobile learning that involve children such as kindergarten students or users with special needs.

Studies that are reluctant to serious mobile learning.

Papers that are only available in the form of abstracts or PowerPoint presentations.

Papers that are not written in English.

Selecting Primary Studies

The titles and abstracts of searched papers were reviewed based on the inclusion and exclusion criteria. Every paper that met at least one of the criteria and without any of the exclusion criteria was included in the review. For papers that could not be excluded based on reading of the titles and abstracts, the full texts of the papers were reviewed. Through this process, 65 articles were selected from the 531 papers initially found. 292 papers were excluded only by reading the topics, 105 papers by reading the abstracts, and 65 papers by reading the full text.

Validation control of the Primary Studies

In order to maintain the quality of the selected studies, the primary studies chosen by the first reviewer were double-checked by a second author. The evaluation of the selected paper was based on the evaluation questions as follows:

Whether a proposed mobile learning solution is implemented in the research context?

Whether the methodology of mobile learning solution is suitable for middle-aged adult?

To what extent the proposed solution effects the middle-aged adult in mobile learning?

The procedure of selecting the primary papers is illustrated in Fig. 2 .

Selecting the primary papers

Data Extraction and Synthesis

In order to extract and synthesize the data to answer the research questions, the selected studies are classified into five categories as follows:

Mobile learning and their research context: This categorization answer the first research question and helps to find the fundamentals and background in mobile research based on research context such as acceptance, adoption, effectiveness, impact, intention of use, usability, and readiness.

Methodology in the mobile learning research area : In order to answer the second research question and find the methodologies employed in the related context, the research context with the methods employed by the researchers was mapped as shown in Table 7 . Based on this mapping, the instruments that have been used in mobile learning research involving middle-aged adults can be identified.

Instruments used in Mobile learning research context: This category answers the second research question. From the systematic mapping done, it was found that the common research instruments used were Questionnaire, Interview, Experiments and Task Analysis. Here, the most preferable instruments used in mobile learning research were highlighted.

Mobile learning solutions in general: This category answers the third research question in order to find the gap in mobile learning research. Articles found in this study include mobile learning articles for young and older adults to show the trend of research towards adulthood. Since the focus of this systematic mapping is on identifying mobile learning technology applied to the middle-aged adults, thus those works focusing on the application of mobile learning not on adult learners or studies on users with special needs were excluded.

Solution for middle-aged adult in mobile learning: This category also answers the third research question in presenting the future works related to mobile learning involving middle-aged adults. This article begins by explaining the use of mobile technology in a learning environment, and the mobile learning theories that form the basis for the comparison of the existing mobile learning solutions for middle-aged adults.

Effects of mobile learning on middle-aged adult: This category answer the importance of the mobile learning towards middle-aged adults for a healthy well-being by assessing the number of studies related to middle-aged adults.

Reporting the Review

In the following section, the outcomes of reviewing the selected studies were reported and the results were discussed in detail, to respond to the defined research questions.

7 Results of the systematic mapping

From the search procedure and criteria, a total number of 65 articles are extracted. The distribution of the primary studies according to the publishing year is shown in Table 4 and Fig. 3 . The articles searched for this systematic review study are from 2016 to 2021. The reason is that this study aims to identify the latest research trends in the field of mobile learning with middle-aged adults. Finding shows that there are several studies from 2016 to 2018 that focus on this topic. The number of articles on mobile learning increased significantly from 2019 to 2020, which may be due to the outbreak of the Covid 19 pandemic. In education, for example, many institutions and organizations have drastically shifted from the traditional teaching and learning approach to online platforms. As a result, there is a considerable amount of research on mobile learning focusing on students in schools, universities, and academic staff. Meanwhile, a lot of study has been done in the field of healthcare with the elderly and middle-aged individuals, because their health begins to decline at this age.

Distribution of reviewed studies by year

It would also be interesting to find out the distribution of studies by countries, as shown in Table 5 . This shows that China had contributed the most research articles in this area of mobile learning. In article [ 13 , 21 , 72 , 77 , 81 , 88 ], the country where the study was conducted was not specified.

8 Participants

The categories of participants in the selected studies consist of young adults, middle-aged adults, and older adults. The number of studies based on age category is illustrated in Table 6 and Fig. 4 . It is found that the number of studies involving young adults is higher compared to studies involving older adults and middle-aged adults. This is due to the fact that young adults are frequent users of smartphones and are more adept at using mobile apps. Furthermore, since they are unable to attend college or universities due to the Covid-19 outbreak, many students are required to study online from home using mobile devices.

Number of studies based on participants’ age category

The details of the reference pertaining to the articles based on participants’ categories (older adults (OA), middle-aged adults (MA), and young adults (YA)) are listed in Table 8 .

9 Research context in Mobile learning

The articles obtained for this study were categorized by research area, as shown in Table 7 . Based on the results, mobile learning was studied in the following areas: Education, Healthcare, Usability, Transactional Services, and Social and Communication. Figure 5 illustrates the number of articles published on each research topic. The finding shows that many researchers prefer to conduct research in the field of education. This is because computers and mobile devices are widely used in educational institutions among young adults. On the other hand, studies that focus on middle-aged and older adults are usually concerned with language or vocabulary learning. The healthcare field is also receiving a lot of attention from researchers, and studies on mobile learning in this field are usually related to elderly and middle-aged people because older people and middle-aged people tend to be more vulnerable to health problems. The number of articles from other fields is low because studies on middle-aged adults and mobile learning did not match the scope and range of years defined for this study.

Number of articles in the research domain

Because the study related to mobile learning is very broad, therefore the article obtained has been classified into research context. Research context was determined based on the previous and current research in the field of mobile learning. It was found that many researchers in the field of mobile learning have studied the acceptance, adoption, effectiveness, impact, intention of use, readiness, and usability of mobile learning. The categorized articles are listed in Table 9 in section 11, with additional information on the methodology used in each study. Figure 6 shows the number of articles obtained by research context.

Number of papers by research context

10 Mobile learning towards the middle-aged adults

From these articles, not many researchers have examined the adoption of mobile learning by middle-aged adults. As mentioned earlier, a person in his or her forties is already inclined to focus on and enhance the standard of daily practice while also finding serenity. At this stage, many people have developed an inclination and willingness to gain more religious knowledge. Adult Muslims who work during the day, would rather choose to visit a mosque or surau to learn about Islam through religious lectures in the evening or at night. During the Covid-19 pandemic outbreak, many people were forced to work from home, and many lectures at the mosque were cancelled. As a result, many have taken the initiative to hold religious lectures via video conferences over the internet (e.g.: Zoom, WebEx). Others tend to watch religious lectures that have been posted on YouTube or other related platforms. However, as opposed to face-to-face seminars, the excitement and encouragement to attend online and prerecorded learning is lacking. Midlife brings with it a multitude of significant life experiences, the majority of which revolve around work, family, especially parenting, and self-development. Tablets are being used more commonly by middle-aged adults to monitor healthy lifestyle behaviors, maintain social contact, avoid injuries, and search information.

Many middle-aged and older adults are using the Internet to obtain information about health conditions and treatments, to get social support and advice from others with similar health-related experiences, and to access apps to help them manage their health [ 28 ]. For instance, Huang, et al. [ 32 ], studied on the attitude of middle-aged adults towards health app usage. From the study, they discovered that middle-aged adults who have no habits in health management tend to consider health applications as valuable tools and have a positive impact on them, while those who already have the habit, do not tend to consider health applications as valuable tool to be used in their daily routines. There are also some middle-aged adults who decide not to use health apps due to some sentimental reasons and the confidence of middle-aged adults in using a smartphone influences their cognitive assessment of health apps.

Table 8 shows the list of studies that are related to middle-aged adults. The age range of the middle-aged adults by each researcher varies. In this study, the age range of the adult is between 40 to 60 years old, which means the selected articles involve participants in this age range. A total of 22 articles were selected that involved middle-aged adults. In the field of language learning, two papers were identified. From these articles, it is found that the study of mobile learning with middle-aged adults is widely conducted in education area. The use of mobile apps in healthcare is also considered important, as this area is also the focus of researchers. The remaining articles are related to the study of user requirements, usability, and the design and development of mobile apps for middle-aged adults.

11 Research methodology

Research methodology is the main key to perform academic research and the strength of a research. The research methodology found used in the selected articles are Questionnaire, Interview, Systematic Literature Review, Literature Review, Reporting, Task Analysis and Experiment. Figure 7 shows the most popular research method used by a researcher in the field of mobile learning is questionnaire (n=24). This methodology has been used in studies that require a large amount of data from many respondents. The second most popular research method used in mobile learning research area is the Interview (n=9). There are also studies that require the use of multiple research methods to answer research questions.

Number of articles by research methodology

Table 9 shows the methodologies employed in the selected articles. However, articles [ 23 , 49 , 63 , 72 , 81 , 83 , 88 ], and [ 77 ] are not included because these articles are review articles.

In the literature, the questionnaire was found to be the most common method used by researchers for data collection involving many participants among young adults and middle-aged adults. On the other hand, the interview method only involved small groups of participants and was carried out in a short time period. Task analysis with interview method was used in three research studies to evaluate the usability, acceptance, and adoption. The studies were done towards young adults and older adults.

In the quantitative research method, the questionnaire instrument was used by the researchers to understand users’ motivation to use e-learning as a medium of learning [ 60 ]; the use of mobile technology and means of internet access [ 24 ]; awareness in using mobile devices towards mobile learning [ 14 ]; investigate the perception of students related to educational use of mobile phones [ 36 , 55 , 76 ]; investigate students’ behavioral intentions [ 3 ] and knowledge transfer among adult workers [ 52 ]; identify factors that affect the intention to use m-learning by learning the experience of the m-learning system by the participants [ 84 ], measure usability [ 19 , 75 ]; use and engagement with m-learning [ 2 ]; collaborative learning experience in social media environment [ 7 ], students’ immersion in the game and their perceived learning outcomes [ 33 ], and the use of mobile application [ 11 , 65 ].

Almost all researchers have formally collected demographic data such as gender, age, degree program, year of study, and race of the participants. There is only one study that collects data on working background because the participants in the study involved working adults. Amongst the selected articles , Al-Adwan, et al. [ 3 ] and Lazar, et al. [ 55 ], validated the content of the survey using experts before the questionnaire was distributed to participants. Dhanapal, et al. [ 19 ] and Huizenga, et al. [ 33 ] carried out a pilot test to identify the flaws and improves the questionnaire. All but one of the researchers used Point Likert scale, while MICAN [ 65 ] uses short answer questions, multiple choices with 1 or n answers, single or two-dimensional questions. The duration of data collection was less than 40 weeks depending on the targeted number of participants.

For the qualitative research method, data was collected via task analysis and interviews. Data were captured through multiple channels including video data analysis and interview content analysis. From the selected articles, it is found that task analysis and interview method were employed in the mobile learning domain to understand participants’ actions, performance, and usability towards mobile apps. The task activities that have been examined by researchers are navigation tasks (with task activity duration of 1.5 hours for older adults to complete searching and navigating using several mobile applications) [ 58 ], quiz activities using Kahoot application (held within 13 weeks for working adult and the task activities were perform in a classroom environment) [ 74 ], mobile devices usage training ( duration of 9 months of training intervention involving older people), and the task activities (e.g.: sending messages, video and audio calls ) was performed in a hospital [ 15 ]; Vocabulary learning [ 86 , 97 ]; games application with task duration of 5 to 20 minutes [ 71 ]; and usability testing [ 30 ]. Open-ended questions were used in the interview sessions [ 71 ] and all the audio recordings of the interviews were transcribed verbatim for analysis purposes [ 58 ].

In the experimental research design, two groups were created with specific condition applied. The treatment group and the control group involved in the experiment and questionnaire research approach can be seen in articles [ 9 , 16 , 38 , 92 ] as listed in Table 9 . For instance, in Bensalem [ 9 ], aims at investigating students' perceptions about the use of WhatsApp in learning vocabulary and in the study, twenty-one participants were randomly assigned to the experimental group. Participants from the experimental group are required to complete and submit their vocabulary assignments via WhatsApp. In the assignment, students are required to search the meaning of new words in a dictionary and build sentences using each word. On the other hand, participants from a control group need to submit the same homework assignment using the traditional paper and pencil method. Later, a questionnaire was distributed to the participants and the collected data was used to measure the participants’ perception of the use of WhatsApp in vocabulary learning.

12 Discussion

In this article, a systematic review was conducted to provide a thorough analysis on the methodologies adopted by researchers in mobile learning. The number of research papers in the year 2020 exceeds the number of research papers in the previous year. This could be due to the outbreak of the Covid-19 pandemic that triggered higher number of papers. During the pandemic, everyone had to work from home, and many organizations, including public and private higher learning institutions, were unable to carry out traditional teaching and learning activities. As a result, many studies or meetings were required to be conducted online.

The country with the highest number of research papers in the field of mobile learning is China with 11 articles. There is a lack of study in mobile learning that focuses on middle-aged adults. Out of 65 research papers, a total of 22 research papers are related to middle-aged adults whereby the distribution of research can be seen in countries such as in Czech Republic (n=1), United States (n=5), China (n=3), Germany (n=1). Singapore (n=2), Turkey (n=1), Brazil (n=1), Poland (n=1), Bangladesh (n=1), United Kingdom (n=2) and 2 articles did not mention the country in which the research was carried out. Studies related to middle-aged adults in Malaysia are not very encouraging, therefore the study of middle-aged adults in the field of mobile learning needs to be given more attention.

The articles selected in this systematic review were classified by research context to identify the focus of previous researchers on the use of mobile learning by middle-aged individuals. Overall, it was found that studies related to the adoption of mobile learning, mobile applications and mobile devices have gained significant attention among researchers, followed by studies related to the acceptance and mobile learning usage. However, studies on examining the adoption and effectiveness of mobile learning usage towards middle-aged adults are still lacking. Examining the effectiveness of mobile learning usage is crucial to provide guidance towards decision making and development work in the future.

The field of education is a popular field for researchers as it involves teachers and young adults who are mainly engaged in the learning environments. Research on middle-aged adults in the educational field is found in seven articles, where two of the articles focused on vocabulary learning. One study on Hadith learning for middle-aged adults, which has been classified as a study on spiritual learning under the educational research domain was also identified. The remaining four articles are respectively related to the use of game applications in teaching adults, the use of mobile devices in sharing information among adult workers, and the readiness of the teachers in adopting mobile learning in a classroom. Besides that, there is also a lack of research towards middle-aged adults in the area of mobile usability and user requirements. Research in the healthcare domain mostly involves older adults where most researchers extensively investigate the use of mobile devices and mobile applications towards healthy ageing and wellbeing.

The coding of the research methods was based on the methods reported by the researchers in their methodology section. Questionnaire is a popular instrument used across quantitative and mixed research approaches for data collection. The questionnaire developed by the researcher will be validated by the experts and tested before it was distributed accordingly to the targeted participants. Task analysis and interview approach can be used to observe the behavior of the users and to evaluate users’ feedback in the learning environment. Even though the method was not extensively used by the researchers from the selected literature focusing on middle-aged adults, this method to be employed in the mobile learning research to gain more insight on the effectiveness of mobile technologies in the learning environment of middle-aged adults was suggested.

Nowadays, almost everyone owns a smartphone, as smartphone prices have dropped significantly, making them affordable for more users. All smartphone users are capable to use most of the basic features of the mobile device, such as downloading applications from the Apple Store or Google Play. Given that middle-aged individuals are heavy smartphone users, it is critical to understand how users utilize mobile technology such as smartphones not just for work, leisure, and entertainment, but also for knowledge acquisition.

Middle-aged adults are self-directed, able to take responsibility for their learning, have a variety of experiences and backgrounds, and are motivated and willing to learn while effectively managing real-world situations. Hence, middle-aged adults can benefit from webinars and short courses delivered online. Therefore, more research should be conducted on mobile learning for middle-aged adults.

13 Conclusion and future work

The novelty of this study is that it contributes to the understanding of the research trends based on research context and methods used in research related to middle-aged adults in mobile learning. It is noted that there are still few studies that address the adoption and effectiveness of mobile apps in the area of religious orientation, especially among middle-aged adults. For instance, before the Covid-19 outbreak, middle-aged Muslims in Malaysia preferred to attend religious courses and trainings to improve their spiritual and religious orientation [ 96 ] based on face-to-face with teachers in a classroom. Therefore, it is critical to determine whether middle-aged adults intend and consent to religious and spiritual learning, such as learning the Quran to be conducted via mobile devices. It is hoped that the use of mobile learning will enable adults' lifelong learning to be improved and done continuously under any situation in the future. This study suggests further studies on middle-aged in the field of mobile learning as follows:

Skills and Knowledge Development

The use of mobile learning among middle-aged adults begins with an awareness and intention to use mobile devices. Generally, middle-aged adults who own smartphones, they already have skills to download apps from the Google Store or App Store and set security preferences. Hence, they must intend to use mobile learning to develop their skills and knowledge. This is because between the ages of 40 and 60, they are usually busy with their work while facing problems such as increasing concerns about health, death of a friend or relative, changes in wages/salaries, and concerns about changes in physical appearance. Therefore, middle-aged adults need to seek knowledge that will make them be satisfied and enable them to lead a better and healthier lifestyle. For example, middle-aged Muslims can learn to understand the Quran through mobile learning to achieve a better quality of life because the Quran is the final revelation and book from Allah s.w.t to humankind as guidance and direction to the right path.

Mobile Learning Application with Multimedia

Mobile learning Application with multimedia plays a great role in motivating learners in learning via digital devices such as smartphones. It is crucial to design and develop mobile learning apps with appropriate multimedia elements such as texts, images, icons, and animations that meet the needs of middle-aged adult learners. In addition, middle-aged adults need to be helped to increase their motivation to learn and improve their memory performance in vocabulary memorization. Therefore, for future work, mobile app development needs to be carefully developed based on user needs especially for the multimedia elements such as the text, graphic, video and animation.

Mobile Learning Application and Quick Assessment

Assessment is a critical component of learning since it demonstrates progress. Because most of the learning occurs online and involves many students, a teacher develops easy assessment tools and procedures that enable them to rapidly assess their students’ learning progress. Numerous game-based apps have aided in the facilitation of teaching and may be used to measure a student learning progress. Additionally, to make mobile learning more interesting, game-based elements have been used to improve the students’ engagement and enjoyment in learning. For instance, Kahoot is a game-based technological platform that can be accessed using, for instance, a smart device or a laptop. The game-based learning application (app) can benefit working adults who are adult learners with diverse learning abilities. Chunking method was used to break down complex concepts into smaller parts in the form of multiple-choice questions. The students’ learning process is tested and corrected, in real time, through the statistics which are generated from this chunking process. Kahoot creates a safe environment for students to make mistakes through multiple choice questions, and yet relearn it without being judged by their peers. However, the drawback of Kahoot is, it does not adequately support the learning experience of adult learners Seah [ 74 ]. Therefore, in the future, the development of mobile learning apps for middle-aged adults might include a gamification aspect that allows easy assessment for self-monitoring of learning progress.

Research Methodology

The finding of this study shows that questionnaire is a popular instrument used across quantitative and mixed research approaches for data collection. The questionnaire developed by the researcher will be validated by the experts and tested before it was distributed accordingly to the targeted participants. However, based on the research context and methodologies found in the literature, the study on middle-aged adults was not getting the enough intention among researchers. Furthermore, as Covid-19 pandemic has impacted people’s life, many are reluctant to participate in answering questionnaires as they may be unmotivated due to job loss, adaptation to new norms or due to the death of their family members. Therefore, in the future, it is hereby recommended that a contribution back to society such as given some tokens to the participants [ 66 , 90 ] can be practiced in the research methodology. Besides that, a researcher also can conduct a free intensive course of related field to a group of respondents to upgrade the lifestyle and well-being among respondents. Hence, this can increase public participation in research, especially when involving busy and elderly respondents and at the same time the respondents can learn new knowledge while also contributing to the research study.

Adedoyin OB, Soykan E (2020) Covid-19 pandemic and online learning: the challenges and opportunities. Interact Learn Environ:1–13

Al Masarweh M (2019) Evaluating M-Learning System Adoption by Faculty Members in Saudi Arabia Using Concern Based Adoption Model (CBAM) Stages of Concern. Int J Emerg Technol Learn 14(5)

Al-Adwan AS, Al-Madadha A, Zvirzdinaite Z (2018) Modeling students’ readiness to adopt mobile learning in higher education: An empirical study. Int Rev Res Open Dist Learn 19(1)

Alkhezzi F, Ahmed MS (2020) A Review of Mobile Learning Technology Integration: Models, Frameworks, and Theories. Coll Stud J 54(4):491–504, Winter2020

Google Scholar

Al-Sabbagh KW, Bradley L, Bartram L (2019) Mobile language learning applications for Arabic speaking migrants–a usability perspective. Lang Learn Higher Educ 9(1):71–95

Article Google Scholar

Amarya S, Singh K, Sabharwal M (2018) Ageing process and physiological changes. In: Gerontology. IntechOpen

Ansari JAN, Khan NA (2020) Exploring the role of social media in collaborative learning the new domain of learning. Smart Learn Environ 7(1):1–16

Anshari M, Almunawar MN, Shahrill M, Wicaksono DK, Huda M (2017) Smartphones usage in the classrooms: learning aid or interference? Educ Inf Technol 22(6):3063–3079

Bensalem E (2018) The Impact of WhatsApp on EFL students' Vocabulary Learning. Arab World Eng J 9(1):23–03

Bere A, Rambe P (2019) Understanding Mobile learning using a social embeddedness approach: a case of instant messaging. Int J Educ Dev Using Inf Commun Technol 15(2):132–153

Cao Y, Qin X, Li J, Long Q, Hu B (2020) Exploring seniors’ continuance intention to use mobile social network sites in China: a cognitive-affective-conative model. Univ Access Inf Soc 21:1–22

Cheng C-H, Chen Y-S, Sangaiah AK, Su Y-H (2018) Evidence-based personal applications of medical computing models in risk factors of cardiovascular disease for the middle-aged and elderly. Pers Ubiquit Comput 22(5):921–936

Chittaro L (2016) Designing serious games for safety education:“learn to brace” Versus traditional pictorials for aircraft passengers. IEEE Trans Vis Comput Graph 22(5):1527–1539

Article MathSciNet Google Scholar

Christensen R, Knezek G (2017) Readiness for integrating mobile learning in the classroom: challenges, preferences and possibilities. Comput Hum Behav 76:112–121. https://doi.org/10.1016/j.chb.2017.07.014

Cid A, Sotelo R, Leguisamo M, Ramírez-Michelena M (2020) Tablets for deeply disadvantaged older adults: challenges in long-term care facilities. Int J Human-Comput Stud 144:102504

Daungcharone K, Panjaburee P, Thongkoo K (2019) A mobile game-based C programming language learning: results of university students' achievement and motivations. Int J Mobile Learn Org 13(2):171–192

de Lara SMA, de Mattos Fortes RP, Russo CM, Freire AP (2016) A study on the acceptance of website interaction aids by older adults. Univ Access Inf Soc 15(3):445–460

Devshikha Bose KP (2020) Lana Grover, "a Mobile learning Community in a Living Learning Community : perceived impact on digital fluency and communication," the online journal of new horizons in education

Dhanapal S, Salman NW, Raman Kutty GK, Ansari KMS, Shah SA (2019) 'XYZ' Application as a Tool for Teaching and Learning in Institutions of Higher Learning: An Exploratory Study. Aplikasi 'XYZ' Sebagai Alat untuk Pengajaran dan Pembelajaran di Institusi Pengajian Tinggi: Satu Kajian Tinjauan (55):1–01

Martino FD, Delmastro F, Dolciotti C (2021) Malnutrition risk assessment in frail older adults using m-health and machine learning. ICC 2021 - IEEE International Conference on Communications, pp 1–6

Ejaz A, Rahim M, Khoja SA (2019) "The Effect of Cognitive Load on Gesture Acceptability of Older Adults in Mobile Application," in 2019 IEEE 10th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON): IEEE, pp. 0979–0986.

Farivar S, Abouzahra M, Ghasemaghaei M (2020) Wearable device adoption among older adults: a mixed-methods study. Int J Inf Manag 55:102209

Galić S, Lušić Z, Stanivuk T (2020) E-LEARNING IN MARITIME AFFAIRS. J Nav Archit Mar Eng 17(1):39, 01//

Gan CL, Balakrishnan V (2018) Mobile Technology in the Classroom: what drives student-lecturer interactions? Int Human-Comput Interact 34(7):666–679. https://doi.org/10.1080/10447318.2017.1380970

García Botero G, Questier F, Zhu C (2019) Self-directed language learning in a mobile-assisted, out-of-class context: do students walk the talk? Comput Assist Lang Learn 32(1–2):71–97

Garg R (2021) "Its changes so often": Parental non-/use of mobile devices while caring for infants and toddlers at home. Proc. ACM Hum.-Comput. Interact., 5(CSCW2), Article 369. https://doi.org/10.1145/3479513

Gong C, Saha K, Chancellor S (2021) "The Smartest Decision for My Future": Social Media Reveals Challenges and Stress During Post-College Life Transition. Proc ACM Human-Comput Interact 5(CSCW2):1–29

Gordon NP, Crouch E (2019) Digital information technology use and patient preferences for internet-based health education modalities: cross-sectional survey study of middle-aged and older adults with chronic health conditions. JMIR aging 2(1):e12243

Grant MM (2019) Difficulties in defining mobile learning: analysis, design characteristics, and implications. Educ Technol Res Dev 67(2):361–388

Guinibert M (2020) Learn from your environment: A visual literacy learning model. Australas J Educ Technol 36(4):173–188. https://doi.org/10.14742/ajet.5200

Hall GS (1922) Senescence, the last half of life. Appleton, New York

Book Google Scholar

Huang Y-M, Lou S-J, Huang T-C, Jeng Y-L (2019) Middle-aged adults’ attitudes toward health app usage: a comparison with the cognitive-affective-conative model. Univ Access Inf Soc 18(4):927–938

Huizenga J, Admiraal W, Ten Dam G, Voogt J (2019) Mobile game-based learning in secondary education: students’ immersion, game activities, team performance and learning outcomes. Comput Hum Behav 99:137–143

Intarasirisawat J, Ang CS, Efstratiou C, Dickens L, Sriburapar N, Sharma D, Asawathaweeboon B (2020) An automated Mobile game-based screening tool for patients with alcohol dependence. Proc ACM Interact, Mobile, Wearab Ubiquitous Technol 4(3):1–23

Introduction to psychology: lifespan development. https://courses.lumenlearning.com/wmopen-psychology/chapter/reading-adulthood/ . Accessed 16 Feb 2021

Iqbal S (2017) Mobile phone usage and students’ perception towards m-learning: A case of undergraduate students in Pakistan. J Distance Educ/Revue de l'ducation Dist 32(1)

Islam MN, Ahmed MA, Islam AN (2020) Chakuri-bazaar: a mobile application for illiterate and semi-literate people for searching employment. Int Mobile Human Comput Interact (IJMHCI) 12(2):22–39

Jeno LM, Dettweiler U, Grytnes J-A (2020) The effects of a goal-framing and need-supportive app on undergraduates' intentions, effort, and achievement in mobile science learning. Comput Educ 159, 12/01/December 2020. https://doi.org/10.1016/j.compedu.2020.104022

Jones SL, Hue W, Kelly RM, Barnett R, Henderson V, Sengupta R (2021) Determinants of longitudinal adherence in smartphone-based self-tracking for chronic health conditions: evidence from axial Spondyloarthritis. Proc ACM Interact, Mobile, Wearab Ubiquitous Technol 5(1):1–24

Jones J, Yuan Y, Yarosh S (2021) Be consistent, work the program, be present every day: exploring Technologies for Self-Tracking in early recovery. Proc ACM Interact, Mobile, Wearab Ubiquitous Technol 5(4):1–26

Justo R et al (2020) Analysis of the interaction between elderly people and a simulated virtual coach. J Ambient Intell Humaniz Comput:1–16

Kadir MA, Arifin S, Latipun, Fuad AN (2016) Adult Learners' understanding in learning Islam using andragogy approach: a study in Kampung Siglap mosque and Al-Zuhri higher learning institute. J Educ Pract 7(32):1–10

Kadirire J, Guy R (2009) Mobile learning demystified. Evol Mobile Teach learn:15–56

Kang H, Lin X (2019) Lifelong learning on the go: English language Mobile learning in China. New Direct Adult Contin Educ 2019(162):49–60

Keegan D (2005)The incorporation of mobile learning into mainstream education and training. In: World Conference on Mobile Learning, Cape Town, vol 11, pp 1–17

Keskin NO, Metcalf D (2011) The current perspectives, theories and practices of mobile learning. Turkish Online J Educ Technol-TOJET 10(2):202–208

Kiconco RI, Rooks G, Snijders C (2020) Learning mobile money in social networks: comparing a rural and urban region in Uganda. Comput Hum Behav 103:214–225

Kitchenham B (2004) Procedures for performing systematic reviews. Keele, UK, Keele University 33(2004):1–26

Klimova B (2020) Benefits of the use of Mobile applications for learning a foreign language by elderly population. Procedia Comput Sci 176:2184–2191

Knowles MS (1975) Self-directed learning: A guide for learners and teachers.

Lippitt GL, Knowles MS, Knowles MS (1984) Andragogy in action: applying modern principles of adult learning

Kuciapski M (2017) A model of mobile technologies acceptance for knowledge transfer by employees. J Knowl Manag 21(5):1053–1076. https://doi.org/10.1108/JKM-03-2016-0136

Lachman ME, Lewkowicz C, Marcus A, Peng Y (1994) Images of midlife development among young, middle-aged, and older adults. J Adult Dev 1(4):201–211

Lachman ME, Teshale S, Agrigoroaei S (2015) Midlife as a pivotal period in the life course: balancing growth and decline at the crossroads of youth and old age. Int J Behav Dev 39(1):20–31

Lazar IM, Panisoara G, Panisoara IO (2020) Digital technology adoption scale in the blended learning context in higher education: Development, validation and testing of a specific tool. PLoS One 15(7):e0235957. https://doi.org/10.1371/journal.pone.0235957

Lee K, Choi H, Cho YH (2019) Becoming a competent self: a developmental process of adult distance learning. Internet High Educ 41:25–33

Lewis R, Taylor-Poleskey M (2021) Hidden town in 3D: teaching and reinterpreting slavery virtually at a living history museum. J Comput Cultural Heritage (JOCCH) 14(2):1–14

Li Q, Luximon Y (2020) Older adults’ use of mobile device: usability challenges while navigating various interfaces. Behav Inform Technol 39(8):837–861

Lin CJ, Ho S-H (2020) The development of a mobile user interface ability evaluation system for the elderly. Appl Ergon 89:103215

Lin X, Su S (2020) Chinese college Students' attitude and intention of adopting Mobile learning. Int J Educ Dev Using Inf Commun Technol 16(2):6–21

Manca M et al (2021) The impact of serious games with humanoid robots on mild cognitive impairment older adults. Int J Hum Comput Stud 145. https://doi.org/10.1016/j.ijhcs.2020.102509

Masrom M, Nadzari AS, Mahmood NHN, Zakaria WNW, Ali NRM (2016) Mobile learning in Malaysia education institutions. Issues in Information Systems 17(4):152-157

Mather CA, Gale F, Cummings EA (2017) Governing mobile technology use for continuing professional development in the Australian nursing profession. BMC Nurs 16(1):1–11

Mendel T, Toch E (2019) My mom was getting this popup: understanding motivations and processes in helping older relatives with mobile security and privacy. Proc ACM Interact, Mobile, Wearab Ubiquitous Technol 3(4):1–20

Mican D (2019) Examining Adoption of Mobile Devices and Applications among College Students in Romania. J App Comput Sci Math 13(28)

Michelson R, DeWitt A, Nagar R, Hiniker A, Yip J, Munson SA, Kientz JA (2021) Parenting in a pandemic: juggling multiple roles and managing technology use in family life during COVID-19 in the United States. In: Proceedings of the ACM on Human-Computer Interaction, 5(CSCW2), pp 1–39

Moore M, DeBruhl B (2019) Investigating university student desires and use of smartphone privacy settings. J Comput Sci Colleges 34(4):134–141

Naismith L, Lonsdale P, Vavoula GN, Sharples M (2004) Literature review in mobile technologies and learning. Future Lab Report, 11

Nordin N, Embi MA, Yunus MM (2010) Mobile learning framework for lifelong learning. Procedia Soc Behav Sci 7:130–138

Ozdamli F, Cavus N (2011) Basic elements and characteristics of mobile learning. Procedia Soc Behav Sci 28:937–942

Palacio RR, Acosta CO, Cortez J, Morán AL (2017) Usability perception of different video game devices in elderly users. Univ Access Inf Soc 16(1):103–113

Palalas A, Wark N (2020) The relationship between mobile learning and self-regulated learning: A systematic review. Australas J Educ Technol 36(4):151–172. https://doi.org/10.14742/ajet.5650

Sarrab M, Baghdadi Y, Al-Shihi H, Bourdoucen H (2016) A model for mobile learning non-functional requirement elicitation. Int J Mobile Learn Org 10(3):129–158

Seah D (2020) Using Kahoot in law school: differentiated instruction for working adults with diverse learning abilities. Int J Mobile Learn Org 14(1):36–48

Sevkli AZ, Motiwalla L, Abdulkarem HF (2017) The design and implementation of a context-aware mobile hadith learning system. Int J Mobile Learn Org 11(4):295–313

Sharma SK, Sarrab M, Al-Shihi H (2017) Development and validation of mobile learning acceptance measure. Interact Learn Environ 25(7):847–858

Shuib L, Ghani NA, Elaish MM, Yadegaridehkordi E (2019) Mobile English language learning (MELL): a literature review. Educ Rev 71(2):257–276. https://doi.org/10.1080/00131911.2017.1382445

Shukla S (2020) M-learning adoption of management students’: a case of India. Educ Inf Technol 26:1–32

Sin F, Berger S, Kim I-J, Yoon D (2021) Digital social interaction in older adults during the COVID-19 pandemic. Proc ACM Human-Comput Interact 5(CSCW2):1–20

Stewart AJ, Ostrove JM, Helson R (2001) Middle aging in women: patterns of personality change from the 30s to the 50s. J Adult Dev 8(1):23–37

Sundgren M (2017) Blurring time and place in higher education with bring your own device applications: a literature review. Educ Inf Technol 22(6):3081–3119

Swanson JA (2020) Assessing the Effectiveness of the Use of Mobile Technology in a Collegiate Course: A Case Study in M-learning. Technol Knowl Learn 25(2):389, 06//

Tan E, Law R (2016) mLearning as a softer visitor management approach for sustainable tourism. J Sustain Tour 24(1):132, 01//

Thongsri N, Shen L, Bao Y, Alharbi IM (2018) "Integrating UTAUT and UGT to explain behavioural intention to use M-learning: a developing country’s perspective," Journal of Systems and Information Technology

Traxler J (2009) Current state of mobile learning. Mobile Learning: Trans del Educ Train 1:9–24

Tu Y, Zou D, Zhang R (2020) A comprehensive framework for designing and evaluating vocabulary learning apps from multiple perspectives. Int J Mobile Learn Org 14(3):370–397

Vacher M, Aman F, Rossato S, Portet F, Lecouteux B (2019) Making emergency calls more accessible to older adults through a hands-free speech interface in the house. ACM Transact Acces Comput (TACCESS) 12(2):1–25

Venkataraman JB, Ramasamy S (2018) Factors influencing mobile learning: a literature review of selected journal papers. Int J Mob Learn Org 12:99–112

Vicente P, Lopes I (2016) Attitudes of older mobile phone users towards mobile phones. Communications 41(1):71–86

Wan Z, Bao L, Gao D, Toch E, Xia X, Mendel T, Lo D (2019) Appmod: helping older adults manage mobile security with online social help. Proc ACM Interact, Mobile, Wearab Ubiquitous Technol 3(4):1–22

Wang Y, Matz-Costa C, Miller J, Carr DC, Kohlbacher F (2018) Uses and gratifications sought from mobile phones and loneliness among japanese midlife and older Adults: a Mediation Analysis. Innov Aging 2(3)

Wardaszko M, Podgórski B (2017) Mobile learning game effectiveness in cognitive learning by adults: a comparative study. Simul Gaming 48(4):435–454

Wong CY, Ibrahim R, Hamid TA, Mansor EI (2018) Usability and design issues of smartphone user interface and mobile apps for older adults. In: International Conference on User Science and Engineering. Springer, pp 93–104

Chapter Google Scholar

Yaffe MJ, Stewart MA (1984) The problems and concerns of middle age. Can Fam Physician 30:1089–1093

Yang H-L, Lin S-L (2019) The reasons why elderly mobile users adopt ubiquitous mobile social service. Comput Hum Behav 93:62–75

Zainal A, Ahmad NA, Razak FHA, Nordin A (2017) Older adults’ requirements of Islamic Mobile applications: a multimethod exploration. Adv Sci Lett 23(5):4236–4240. https://doi.org/10.1166/asl.2017.8335

Zhang R, Zou D (2020) Influential factors of working adults' perceptions of mobile-assisted vocabulary learning with multimedia annotations. Int J Mobile Learn Org 14(4):533–548

Download references

Acknowledgments

We would like to thank you to Universiti Teknikal Malaysia Melaka (UTeM) for sponsoring the corresponding author in this research numbered UTeM.02.13.04/500-4/12/16/1/2(91)

Conflict of interest

The authors declare that there is no conflict of interest regarding the publication of this paper.

Availability of data and materials

The datasets analyzed during the current study are available from the corresponding author on reasonable request.

The corresponding author is sponsored by Universiti Teknikal Malaysia Melaka (UTeM) numbered UTeM.02.13.04/500-4/12/16/1/2(91) to pursue her PhD degree in these studies.

Author information

Authors and affiliations.

Department of Software Engineering, Faculty of Computer Science and Information Technology, Universiti Malaya, 50603, Kuala Lumpur, Malaysia

Syahida Mohtar, Nazean Jomhari & Mumtaz Begum Mustafa

Department of Al-Quran and Al-Hadith, Academy of Islamic Studies, Universiti Malaya, 50603, Kuala Lumpur, Malaysia

Zulkifli Mohd Yusoff

You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Syahida Mohtar .

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

Mohtar, S., Jomhari, N., Mustafa, M.B. et al. Mobile learning: research context, methodologies and future works towards middle-aged adults – a systematic literature review. Multimed Tools Appl (2022). https://doi.org/10.1007/s11042-022-13698-y

Download citation

Received : 02 August 2021

Revised : 17 June 2022

Accepted : 15 August 2022

Published : 20 August 2022

DOI : https://doi.org/10.1007/s11042-022-13698-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Mobile application
Self-directed learning
Middle-aged adults
Methodology
Find a journal
Publish with us
Track your research

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

View all journals
My Account Login
Explore content
About the journal
Publish with us
Sign up for alerts
Open access
Published: 30 April 2024

Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b

Taylor J. Bell ORCID: orcid.org/0000-0003-4177-2149 1 , 2 ,
Nicolas Crouzet ORCID: orcid.org/0000-0001-7866-8738 3 ,
Patricio E. Cubillos 4 , 5 ,
Laura Kreidberg 6 ,
Anjali A. A. Piette ORCID: orcid.org/0000-0002-4487-5533 7 ,
Michael T. Roman ORCID: orcid.org/0000-0001-8206-2165 8 , 9 ,
Joanna K. Barstow ORCID: orcid.org/0000-0003-3726-5419 10 ,
Jasmina Blecic 11 , 12 ,
Ludmila Carone ORCID: orcid.org/0000-0001-9355-3752 5 ,
Louis-Philippe Coulombe ORCID: orcid.org/0000-0002-2195-735X 13 ,
Elsa Ducrot ORCID: orcid.org/0000-0002-7008-6888 14 ,
Mark Hammond 15 ,
João M. Mendonça 16 ,
Julianne I. Moses ORCID: orcid.org/0000-0002-8837-0035 17 ,
Vivien Parmentier 18 ,
Kevin B. Stevenson ORCID: orcid.org/0000-0002-7352-7941 19 ,
Lucas Teinturier ORCID: orcid.org/0000-0002-0797-5746 20 , 21 ,
Michael Zhang 22 ,
Natalie M. Batalha 23 ,
Jacob L. Bean 22 ,
Björn Benneke ORCID: orcid.org/0000-0001-5578-1498 13 ,
Benjamin Charnay 20 ,
Katy L. Chubb 24 ,
Brice-Olivier Demory 25 , 26 ,
Peter Gao 7 ,
Elspeth K. H. Lee 25 ,
Mercedes López-Morales 27 ,
Giuseppe Morello ORCID: orcid.org/0000-0002-4262-5661 28 , 29 , 30 ,
Emily Rauscher ORCID: orcid.org/0000-0003-3963-9672 31 ,
David K. Sing ORCID: orcid.org/0000-0001-6050-7645 32 , 33 ,
Xianyu Tan 15 , 34 , 35 ,
Olivia Venot ORCID: orcid.org/0000-0003-2854-765X 36 ,
Hannah R. Wakeford ORCID: orcid.org/0000-0003-4328-3867 37 ,
Keshav Aggarwal ORCID: orcid.org/0000-0002-7004-8670 38 ,
Eva-Maria Ahrer 39 , 40 ,
Munazza K. Alam ORCID: orcid.org/0000-0003-4157-832X 7 ,
Robin Baeyens ORCID: orcid.org/0000-0001-7578-969X 41 ,
David Barrado ORCID: orcid.org/0000-0002-5971-9242 42 ,
Claudio Caceres ORCID: orcid.org/0000-0002-6617-3823 43 , 44 , 45 ,
Aarynn L. Carter 23 ,
Sarah L. Casewell 8 ,
Ryan C. Challener 31 ,
Ian J. M. Crossfield 46 ,
Leen Decin ORCID: orcid.org/0000-0002-5342-8612 47 ,
Jean-Michel Désert 41 ,
Ian Dobbs-Dixon 11 ,
Achrène Dyrek 14 ,
Néstor Espinoza 33 , 48 ,
Adina D. Feinstein 22 , 49 ,
Neale P. Gibson 50 ,
Joseph Harrington ORCID: orcid.org/0000-0002-8955-8531 51 ,
Christiane Helling 5 ,
Renyu Hu ORCID: orcid.org/0000-0003-2215-8485 52 , 53 ,
Nicolas Iro 54 ,
Eliza M.-R. Kempton ORCID: orcid.org/0000-0002-1337-9051 55 ,
Sarah Kendrew 56 ,
Thaddeus D. Komacek ORCID: orcid.org/0000-0002-9258-5311 55 ,
Jessica Krick 57 ,
Pierre-Olivier Lagage 14 ,
Jérémy Leconte 58 ,
Monika Lendl ORCID: orcid.org/0000-0001-9699-1459 59 ,
Neil T. Lewis 60 ,
Joshua D. Lothringer 61 ,
Isaac Malsky ORCID: orcid.org/0000-0003-0217-3880 31 ,
Luigi Mancini ORCID: orcid.org/0000-0002-9428-8732 6 , 62 , 63 ,
Megan Mansfield ORCID: orcid.org/0000-0003-4241-7413 64 ,
Nathan J. Mayne ORCID: orcid.org/0000-0001-6707-4563 65 ,
Thomas M. Evans-Soma ORCID: orcid.org/0000-0001-5442-1300 6 , 66 ,
Karan Molaverdikhani ORCID: orcid.org/0000-0002-0502-0428 67 , 68 ,
Nikolay K. Nikolov ORCID: orcid.org/0000-0002-6500-3574 48 ,
Matthew C. Nixon ORCID: orcid.org/0000-0001-8236-5553 55 ,
Enric Palle ORCID: orcid.org/0000-0003-0987-1593 28 ,
Dominique J. M. Petit dit de la Roche 59 ,
Caroline Piaulet ORCID: orcid.org/0000-0002-2875-917X 13 ,
Diana Powell 27 ,
Benjamin V. Rackham ORCID: orcid.org/0000-0002-3627-1676 69 , 70 ,
Aaron D. Schneider ORCID: orcid.org/0000-0002-1448-0303 47 , 71 ,
Maria E. Steinrueck ORCID: orcid.org/0000-0001-8342-1895 6 ,
Jake Taylor 13 , 15 ,
Luis Welbanks ORCID: orcid.org/0000-0003-0156-4564 72 ,
Sergei N. Yurchenko ORCID: orcid.org/0000-0001-9286-9501 73 ,
Xi Zhang ORCID: orcid.org/0000-0002-8706-6963 74 &
Sebastian Zieba ORCID: orcid.org/0000-0003-0562-6750 3 , 6

Nature Astronomy ( 2024 ) Cite this article

2052 Accesses

528 Altmetric

Metrics details

Hot Jupiters are among the best-studied exoplanets, but it is still poorly understood how their chemical composition and cloud properties vary with longitude. Theoretical models predict that clouds may condense on the nightside and that molecular abundances can be driven out of equilibrium by zonal winds. Here we report a phase-resolved emission spectrum of the hot Jupiter WASP-43b measured from 5 μm to 12 μm with the JWST’s Mid-Infrared Instrument. The spectra reveal a large day–night temperature contrast (with average brightness temperatures of 1,524 ± 35 K and 863 ± 23 K, respectively) and evidence for water absorption at all orbital phases. Comparisons with three-dimensional atmospheric models show that both the phase-curve shape and emission spectra strongly suggest the presence of nightside clouds that become optically thick to thermal emission at pressures greater than ~100 mbar. The dayside is consistent with a cloudless atmosphere above the mid-infrared photosphere. Contrary to expectations from equilibrium chemistry but consistent with disequilibrium kinetics models, methane is not detected on the nightside (2 σ upper limit of 1–6 ppm, depending on model assumptions). Our results provide strong evidence that the atmosphere of WASP-43b is shaped by disequilibrium processes and provide new insights into the properties of the planet’s nightside clouds. However, the remaining discrepancies between our observations and our predictive atmospheric models emphasize the importance of further exploring the effects of clouds and disequilibrium chemistry in numerical models.

Diurnal variations in the stratosphere of the ultrahot giant exoplanet WASP-121b

A broadband thermal emission spectrum of the ultra-hot Jupiter WASP-18b

UV absorption by silicate cloud precursors in ultra-hot Jupiter WASP-178b

Hot Jupiters are tidally synchronized to their host stars, with vast differences in irradiation between the dayside and nightside. Previous observations with the Hubble Space Telescope (HST) and the Spitzer Space Telescope show that these planets have cooler nightsides and weaker hotspot offsets than expected from cloud-free three-dimensional models 1 , 2 , 3 , 4 , 5 . The main mechanism believed to be responsible for this behaviour is the presence of nightside clouds, which would hide the thermal flux of the planet and lead to a sharp longitudinal gradient in brightness temperature 3 , 4 , 6 , 7 , 8 , 9 , 10 . Other mechanisms have been proposed, such as the presence of atmospheric drag due to hydrodynamic instabilities or magnetic coupling 11 , 12 , 13 , super-stellar atmospheric metallicity 14 , 15 , or interaction between the deep winds and the photosphere 16 , but these mechanisms are less universal than the cloud hypothesis 17 , 18 .

WASP-43b, a hot Jupiter with an orbital period of just 19.5 h (ref. 19 ), is an ideal target for thermal phase-curve observations. Its host star is a K7 main-sequence star 87 pc away with metallicity close to solar and weak variability 20 . Previous measurements of the planet’s orbital phase curve in the near-infrared have revealed a large temperature contrast between the dayside and nightside hemispheres, broadly consistent with the presence of nightside clouds 3 , 21 , 22 , which could be composed of magnesium silicates (Mg 2 SiO 4 /MgSiO 2 ) and other minerals (for example, MnS, Na 2 S, metal oxides) 23 , 24 . Owing to the low nightside flux, the exact temperature and cloud properties were challenging to determine from previous observations 4 , 25 , 26 . With the mid-infrared capabilities of the JWST, we have the opportunity to measure the phase-resolved thermal spectrum with unprecedented sensitivity, particularly on the cold nightside. We observed a full orbit of WASP-43b in the 5–12 μm range with the JWST’s Mid-Infrared Instrument (MIRI) 27 in low-resolution spectroscopy (LRS) 28 slitless mode on 1 and 2 December 2022, as part of the Transiting Exoplanet Community Early Release Science Program (JWST-ERS-1366). This continuous observation lasted 26.5 h at a cadence of 10.34 s (9,216 integrations) and included a full phase curve with one transit and two eclipses.

We performed multiple independent reductions and fits to these observations (see ‘Data reduction pipelines’ and ‘Light-curve fitting’ in Methods ) to ensure robust conclusions. Our analyses all identified a strong systematic noise feature from 10.6 μm to 11.8 μm, the source of which is still unclear, and we were unable to adequately detrend these 10.6–11.8 μm data (see ‘Shadowed region effect’ in Methods ). As shown in Extended Data Fig. 1 , we also found that larger wavelength bins were required to accurately estimate our final spectral uncertainties (see ‘Spectral binning’ in Methods ). As a result, our final analyses consider only the 5–10.5 μm data, which we split into 11 channels with a constant 0.5 μm wavelength spacing. Similar to the MIRI commissioning time-series observations, our data show a strong downwards exponential ramp in the first ~60 min and a weaker ramp throughout the observation 29 (Extended Data Fig. 2 ). To minimize correlations with the phase variations, we removed the initial strong ramp by excluding the first 779 integrations (134.2 min) and then fitted a single exponential ramp model to the remaining data. A single ramp effectively removed the systematic noise, with the broadband light curve showing scatter ~1.25× the expected photon noise, while the spectroscopic light curves reach as low as ~1.1× the photon limit, probably due to improved decorrelation of wavelength-dependent systematics. Figure 1 shows the spectral light curves, broadband light curve, dayside spectra and nightside spectra from our fiducial reduction and fit.

a , The observed spectroscopic light curves binned to a 0.5 μm wavelength resolution and after systematic noise removal, following the Eureka! v1 methods. The first 779 integrations have been removed from this figure and our fits as they were impacted by strongly decreasing flux. Wavelengths longer than 10.5 μm marked with a hatched region were affected by the ‘shadowed region effect’ ( Methods ) and could not be reliably reduced. b , The observed band-integrated light curve after systematic noise removal (grey points) and binned data with a cadence of 15 min (black points, with error bars smaller than the point sizes), compared with the best-fitting astrophysical model (red line). c , d , The measured dayside ( c ) and nightside ( d ) emission spectra are shown with black points and 1 σ error bars, and black-body curves (dotted line denoted as ‘BB’, assuming a PHOENIX 74 , 75 , 76 model for the star) are shown to emphasize planetary spectral features with black-body temperatures estimated by eye to match the continuum flux levels. Wavelengths longer than 10.5 μm were affected by the shadowed region effect and are unreliable.

From our Eureka! v1 analysis ( Methods ), we measure a broadband (5–10.5 μm) peak-to-trough phase variation of 4,180 ± 33 ppm with an eclipse depth of 5,752 ± 19 ppm and a nightside flux of 1,636 ± 37 ppm. Assuming a PHOENIX stellar model and marginalizing over the published stellar and system parameters 30 , the broadband dayside brightness temperature is 1,524 ± 35 K while the nightside is 863 ± 23 K. This corresponds to a day–night brightness temperature contrast of 659 ± 19 K, in agreement with the large contrasts previously observed 4 , 21 , 22 , 25 . The phase variations are well fitted by a sum of two sinusoids (the first and second harmonics), with two sinusoids preferred over a single sinusoid at 16 σ (see ‘Determining the number of sinusoid harmonics’ in Methods ) for the broadband light curve. The peak brightness of the broadband phase curve occurs at 7.34 ± 0.38° E from the substellar point (although individual reductions find offsets ranging from 7.34° E to 9.60° E), while previous studies have found offsets of 12.3 ± 1.0° E for HST Wide Field Camera 3’s (WFC3) 1.1–1.7 μm bandpass 21 , offsets ranging from 4.4° E to 12.2° E for Spitzer InfraRed Array Camera’s (IRAC) 3.6 μm filter 22 , 25 and offsets ranging from 10.4° E to 21.1° E for Spitzer/IRAC’s 4.5 μm filter 4 , 22 , 25 , 26 , 31 , 32 . Overall, these broadband data represent roughly an order of magnitude in improved precision on the eclipse depth (6×), phase-curve amplitude (6×) and phase-curve offset (10×) over individual Spitzer/IRAC 4.5 μm observations of the system 22 , 26 , 32 ; this improvement is largely driven by the JWST’s larger mirror (45×), about 12× less pointing jitter (per axis), about 4× improved stability in the width of the point spread function (PSF) along each axis and MIRI’s much broader bandpass.

Model interpretation

To interpret the measurements, we compared the observations with synthetic phase curves and emission spectra derived from general circulation models (GCMs). Simulations were gathered from five different modelling groups, amounting to 31 separate GCM realizations exploring a range of approaches and assumptions. Notably, in addition to cloud-free simulations, the majority of the GCMs modelled clouds with spatial distributions that were either fully predicted 5 , 26 , 33 or simply limited to the planet’s nightside 4 . For the predictive cloud models, simulations favoured warmer, clearer daysides with cooler, cloudier nightsides, but the precise distributions varied with assumptions regarding cloud physics and compositions. In general, models with smaller cloud particles or extended vertical distributions tended to produce thicker clouds at the pressures sensed by the observations. Details of the different models are provided in Methods .

Despite fundamental differences in the models and the parameterizations they employ, simulated phase curves derived from models that include cloud opacity on the planet’s nightside provide a better match to the observed nightside flux compared with the clear simulations (Fig. 2 ). In contrast, the observed dayside fluxes (180° orbital phase) were matched similarly well by models with and without clouds. This implies the presence of widespread clouds preferentially on the planet’s nightside with cloud optical thicknesses sufficient to suppress thermal emission and cool the thermal photosphere. Specifically, models with integrated mid-infrared cloud opacities of roughly 2–4 above the 300 mbar level (that is, blocking ~87–98% of the underlying emission), best match the observed nightside flux.

The black points show the temporally binned broadband light curve. The solid lines represent modelled phase curves derived from the 31 GCM simulations, integrated over the same wavelength range as the data, and separated into two groups based on the inclusion of clouds. The cloudless GCMs (red lines) simulated completely cloud-free skies, whereas the cloudy GCMs (blue lines) included at least some clouds on the nightside of the planet. The red and blue shaded areas span the range of all the cloudless and cloudy simulations, respectively, with the spread of values owing to differences in the various model assumptions and parameterizations. On average, the cloudless GCM phase curves have a maximum planet-to-star flux ratio of 5,703 ppm and a minimum of 2,681 ppm. This matches the observed maximum of the phase curve well but does not match its observed minimum at 1,636 ± 37 ppm. On average, the cloudy GCM phase curves have a maximum of 5,866 ppm and a minimum of 1,201 ppm, in better agreement with the observed nightside emission, but their spread of maximum values is much larger than the cloudless simulations. The cloudy models are able to suppress the nightside emission and better match the data; however, not all cloud models fit equally well and those with the optically thickest nightside clouds suppress too much emission. The models do not include the eclipse signals (phases −0.5 and 0.5) or transit signal (phase 0.0).

Including nightside clouds also improved the agreement with the measured hotspot offset (7.34± 0.38° E). While cloudless models all produced eastward offsets greater than 16.6° (25.5° on average), simulations with clouds had offsets as low as 7.6° (with a mean of 16.4°). These reduced offsets were associated with decreases in the eastwards jet speeds of up to several kilometres per second, with maximum winds of roughly 2.0–2.5 km s −1 providing the best match (see Extended Data Table 1 for further details). This modelled jet-speed reduction is probably due to a disruption in the equatorwards momentum transport 34 brought about by nightside clouds 4 , 35 , 36 . However, the resulting range of offsets seen in the suite of models suggests that this mechanism is quite sensitive to the details of cloud models, and other modelling factors (for example, atmospheric drag 11 , 12 , 16 , radiative timescales 14 , 15 , 37 ) probably still play an important role.

A comparison of the observed and modelled emission spectra further suggests that the majority of the cloud thermal opacity must be confined to pressures greater than ~10–100 mbar, because the presence of substantial cloud opacity at lower pressures dampens the modelled spectral signature amplitude below what is observed (Fig. 3 ). No distinct spectral signatures indicative of the cloud composition were evident in the observations. While no single GCM can match the emission spectra at all phases, spectra corresponding to nightside, morning and evening terminators appear qualitatively similar to GCM results that are intermediate between clear and cloudy simulations. In contrast, the absorption features indicative of water vapour (between ~5 μm and 8.5 μm) seen in the dayside emission spectrum are more consistent with an absence of cloud opacity at these mid-infrared wavelengths. Altogether, these findings represent new constraints on the spatial distribution and opacity of WASP-43b’s clouds.

a – c , The observed emission spectrum with 1 σ error bars at phases 0.0 ( a ), 0.25 ( b ), 0.5 ( c ) and 0.75 ( d ), along with select modelled spectra derived from different cloudy and cloudless GCMs (described in Methods and listed in Extended Data Table 1 ). Although absolute brightness temperatures differ appreciably between models owing to various GCM assumptions, differences in the relative shape of the spectra are strongly dependent on the cloud and temperature structure found in the GCMs (Extended Data Fig. 7 ). Models with more isothermal profiles (like RM-GCM) or thick clouds at pressures of ≲ 10–100 mbar (like THOR cloudy, Generic PCM with 0.1 μm cloud particles) produce flatter spectra, while clearer skies yield stronger absorption features. The observed spectra from the nightside and terminators appear muted compared with the clear-model spectra, suggesting the presence of at least some clouds or weak vertical temperature gradients at pressures of ≲ 10–100 mbar. In contrast, the spectral structure produced by water vapour opacity (indicated by the purple shading) appears more consistent with models lacking clouds at these low pressures on the dayside. Under equilibrium chemistry, methane would also show an absorption feature at ~7.5–8.5 μm (shaded pink) for the colder models at phases 0.0 and 0.75. Finally, the median retrieved spectrum and 1 σ contours from the HyDRA retrieval are shown in grey.

We further characterized the chemical composition of WASP-43b’s atmosphere by applying a suite of atmospheric retrieval frameworks to the phase-resolved emission spectra. The retrievals spanned a broad range of model assumptions, including free chemical abundances versus equilibrium chemistry, different temperature profile parameterizations and different cloud models (see ‘Atmospheric retrieval models’ in Methods ). Despite these differences, the retrievals yielded consistent results for both the chemical and thermal constraints. We detected water vapour across the dayside, nightside, morning and evening hemispheres, with detection significances of up to ~3–4 σ (Extended Data Fig. 3 and Extended Data Tables 2 and 3 ). The retrieved abundances of H 2 O largely lie in the 10–10 5 ppm range for all four phases and for all the retrieval frameworks (Fig. 4 and Extended Data Fig. 4 ), broadly consistent with the value expected for a solar composition (500 ppm) as well as previous observations 22 .

a , Temperature profile contours (68% confidence) constrained by the retrievals at each orbital phase (see legends). All frameworks produced consistent non-inverted thermal profiles that are consistent with two-dimensional radiative–convective equilibrium and photochemical models along the equator 23 (black curves) over the range of pressures probed by the observations (black bars). b , H 2 O abundance posterior distributions (volume mixing ratios). The shaded areas denote the span of the 68% confidence intervals. The green and blue bars on each panel denote the abundances predicted by equilibrium and disequilibrium chemistry solar-abundance models 23 , respectively, at the pressures probed by the observations (1–10 −3 bar, approximately). c , The same as in b but for CH 4 . The retrieved water abundances are consistent with either equilibrium or disequilibrium chemistry estimations for solar composition (500 ppm), whereas the retrieved upper limits to the CH 4 abundance are more consistent with disequilibrium chemistry predictions.

We also searched for signatures of disequilibrium chemistry in the atmosphere of WASP-43b. While CH 4 is expected to be present on the nightside under thermochemical equilibrium conditions, we did not detect CH 4 at any phase (Fig. 4 ). In the pressure range probed by the nightside spectrum (1–10 −3 bar; Extended Data Fig. 5 ), the equilibrium abundance of CH 4 is expected to vary between ~1 ppm and 100 ppm for a solar C/O ratio 23 , compared with our 95% upper limits of 1–6 ppm (Extended Data Table 2 ). The upper limits we place on the nightside CH 4 abundance are more consistent with disequilibrium models that account for vertical and horizontal transport 23 , 24 , 38 . In particular, two-dimensional photochemical models and GCMs predict the strongest depletion of CH 4 on the nightside due to strong zonal winds (>1 km s −1 ) transporting gas-phase constituents around the planet faster than the chemical reactions can maintain thermochemical equilibrium, thus ‘quenching’ and homogenizing the global composition at values more representative of dayside conditions (see also refs. 39 , 40 , 41 , 42 ). We note, however, that a low atmospheric C/O ratio and/or clouds at photospheric pressures could also lead to a non-detection of CH 4 . We also searched for signatures of NH 3 , which is predicted to have a volume mixing ratio less than 0.1–1 ppm in both equilibrium and disequilibrium chemistry models, and find that the results are inconclusive and model-dependent with the current retrieval frameworks.

Given the strong evidence for clouds from comparison with GCMs, we also searched for signatures of clouds in the atmospheric retrieval. Formally, the retrievals do not detect clouds with statistical significance, indicating that strong spectral features uniquely attributable to condensates are not visible in the data (see ‘Atmospheric retrieval models’ in Methods and Extended Data Fig. 6 ). However, the retrievals may mimic the effects of cloud opacity with a more isothermal temperature profile, as both tend to decrease the amplitude of spectral features, but the cloud-free, more isothermal temperature profile requires fewer free parameters and is therefore statistically favoured. Indeed, while the retrieved temperature profiles on the dayside and evening hemispheres agree well with the hemispherically averaged temperature profiles from the GCMs, they are more isothermal than the GCM predictions for the nightside and morning hemispheres (Extended Data Fig. 7 ). This discrepancy may hint at the presence of clouds on the nightside and morning hemispheres, consistent with the locations of clouds found in the GCMs.

Taken together, our results highlight the unique capabilities of JWST/MIRI for exoplanet atmosphere characterization. Combined with a range of atmospheric models, the observed phase curve and emission spectra provide strong evidence that the atmospheric chemistry of WASP-43b is shaped by complex disequilibrium processes and provide new constraints on the optical thickness and pressure of nightside clouds. However, while cloudy GCM predictions match the data better than cloud-free models, none of the simulations simultaneously reproduced the observed phase curve and spectra within measured uncertainties. These remaining discrepancies underscore the importance of further exploring the effects of clouds and disequilibrium chemistry in numerical models, as JWST continues to place unprecedented observational constraints on smaller and cooler planets.

Observations and quality of the data

We observed a full orbit of WASP-43b with the JWST MIRI LRS slitless mode as a part of JWST-ERS-1366. We performed target acquisition with the F1500W filter and used the SLITLESSPRISM subarray for the science observation. The science observation was taken between 1 December 2022 at 00:54:30 UT and 2 December 2022 at 03:23:36 UT, for a total of 26.5 h. We acquired 9,216 integrations, which were split into 3 exposures and 10 segments per exposure. Each integration lasts 10.34 s and is composed of 64 groups, with 1 frame per group. The LRS slitless mode reads an array of 416 × 72 pixels on the detector (the SLITLESSPRISM subarray) and uses the FASTR1 readout mode, which introduces an additional reset between integrations.

Owing to the long duration of the observation, two high-gain antenna moves occurred 8.828 h and 17.661 h after the start of the science observation. They affect only a couple of integrations that we removed from the light curves. A cross-shaped artefact is present on the two-dimensional images at the short-wavelength end due to light scattered by detector pixels 43 . It is stable over the duration of the observation but it contaminates the background and the spectral trace up to ~6 μm. This ‘cruciform’ artefact is observed in all MIRI LRS observations; a dedicated analysis is underway to estimate and mitigate its impact.

In the broadband light curve, the flux decays by ~0.1% during the first 60 min and continues to decay throughout the observation. This ramp is well modelled with 1 or 2 exponential functions after trimming the initial ~780 integrations. Without trimming any data, at least two ramps are needed. In addition, a downwards linear trend in flux is observed over the whole observation with a slope of −39 ppm per hour. These two types of drift also appear in the spectroscopic light curves. The exponential ramp amplitude in the first 60 min changes with wavelength from −0.67% in the 5–5.5 μm bin (downwards ramp) to +0.26% in the 10–10.5 μm bin (upwards ramp). The ramp becomes upwards at wavelengths longer than 7.5 μm and its timescale increases to more than 1 h at wavelengths longer than 10.5 μm. The slopes as a function of wavelength vary from −16 ppm to −52 ppm, all downwards. Such drifts (initial ramp and linear or polynomial trend) are also observed in other MIRI LRS time-series observations 29 but the strength of the trends differ for each observation. In these WASP-43b observations, we note that their characteristic parameters vary smoothly with wavelength, which may help identify their cause and build correction functions.

Over the course of the observation, the position of the spectral trace on the detector varies by 0.0036 pixels RMS (0.027 pixels peak to peak) in the spatial direction, and the Gaussian standard deviation of the spatial PSF varies by 0.00069 pixels RMS (root mean square; 0.0084 pixels peak to peak) following a sharp increase by 0.022 pixels during the first 600 integrations. Depending on the wavelength bin, that spatial drift causes noise at the level of 7–156 ppm, while variations in the PSF width cause noise at the level of 4–54 ppm (these numbers are obtained from a linear decorrelation). Overall, the MIRI instrument used in LRS slitless mode remains remarkably stable over this 26.5-h-long continuous observation and the data are of exquisite quality.

The noise in the light curve increases sharply at wavelengths beyond 10.5 μm and the transit depths obtained at these long wavelengths by different reduction pipelines are discrepant. These wavelengths were not used in the retrieval analyses and the final broadband light curve. The cause is unknown but it might be related to the fact that this region of the detector receives different illumination before the observation 44 (see ‘Shadowed region effect’ below for more details).

Data reduction pipelines

Eureka v1 reduction.

The Eureka! v1 reduction made use of version 0.9 of the Eureka! pipeline 45 , CRDS version 11.16.16 and context 1018, and jwst package version 1.8.3 46 . The gain value of 5.5 electrons per data number obtained from these CRDS reference files is known to be incorrect, and the actual gain is estimated to be ~3.1 electrons per data number although the gain may be wavelength dependent (S. Kendrew, private communication). A new reference file reflecting the updated gain is under development at STScI, which will improve the accuracy of photon-noise calculations. For the rest of this analysis, we assume a constant gain of 3.1 electrons per data number. The Eureka! control files and Eureka! parameter files files used in these analyses are available for download ( https://doi.org/10.5281/zenodo.10525170 ) and are summarized below.

Eureka! makes use of the jwst pipeline for stages 1 and 2, and both stages were run with their default settings, with the exception of increasing the stage 1 jump step’s rejection threshold to 8.0 and skipping the photom step in stage 2 because it is not necessary and can introduce additional noise for relative time-series observations. In stage 3 of Eureka!, we then rotated the MIRI/LRS slitless spectra 90° anticlockwise so that wavelength increases from left to right like the other JWST instruments to allow for easier reuse of Eureka! functions. We then extracted pixels 11–61 in the new y direction (the spatial direction) and 140–393 in the new x direction (spectral direction); pixels outside of these ranges primarily contain noise that is not useful for our reduction. Pixels marked as ‘DO_NOT_USE’ in the DQ array were then masked as were any other unflagged NaN or inf pixels. A centroid was then fit to each integration by summing along the spectral direction and fitting the resulting one-dimensional profile with a Gaussian function; the centroid from the first integration was used for determining aperture locations, while the centroids and PSF widths from all integrations were saved to be used as covariates when fitting the observations.

Our background subtraction method is tailored to mitigate several systematic effects unique to the MIRI instrument. First, MIRI/LRS observations exhibit a ‘cruciform artefact’ 43 at short wavelengths caused by scattered light within the optics; this causes bright rays of scattered light which must be sigma-clipped to avoid over-subtracting the background. In addition, MIRI/LRS observations show periodic noise in the background flux, which drifts with time 29 as well as 1/ f noise 47 , which leads to correlated noise in the cross-dispersion direction; as a result, background subtraction must be performed independently for each integration and column (row in MIRI’s rotated reference frame). Furthermore, in both these observations and the dedicated background calibration observations from JWST-COM/MIRI-1053, we found that there was a linear trend in the background flux, with the background flux increasing with increasing row index (column index in MIRI’s rotated reference frame). To robustly remove this feature, we found that it was important to either (1) use the mean from an equal number of pixels on either side of the spectral trace for each column and integration, or (2) use a linear background model for each column and integration; we adopted the former as it resulted in less noisy light curves. To summarize, for each column in each integration we subtracted the mean of the pixels separated by ≥11 pixels from the centre of the spectral trace after first masking 5 σ outliers in that column.

To compute the spatial profile for the optimal extraction of the source flux, we calculated a median frame, sigma-clipping 5 σ outliers along the time axis and smoothing along the spectral direction using a 7-pixel-wide boxcar filter. During optimal extraction, we only used the pixels within 5 pixels of the fitted centroid and masked pixels that were 10 σ discrepant with the spatial profile. Background exclusion regions ranging from 9 to 13 pixels and source aperture regions ranging from 4 to 6 pixels were considered, but our values of 11 and 5 pixels were selected as they produced the lowest median absolute deviation light curves before fitting.

Eureka! v2 reduction

The Eureka! v2 reduction followed the same procedure as the Eureka! v1 reduction except for the following differences. First, this reduction made use of version 1.8.1 of the jwst pipeline. For stage 1, we instead used a cosmic ray detection threshold of 5 and used a uniform ramp fitting weighting. For stage 2, we performed background subtraction using columns away from the trace on the left and on the right and subtracted the background for each integration 29 . Stage 3 was identical to Eureka! v1 reduction.

TEATRO reduction

We processed the data using the Transiting Exoplanet Atmosphere Tool for Reduction of Observations (TEATRO) that runs the jwst package, extracts and cleans the stellar spectra and light curves, and runs light-curve fits. We used the jwst package version 1.8.4, CRDS version 11.16.14 and context 1019. We started from the ‘uncal’ files and ran stages 1 and 2 of the pipeline. For stage 1, we set a jump rejection threshold of 6, turned off the ‘jump.flag_4_neighbors’ parameter and used the default values for all other parameters. For stage 2, we ran only the ‘AssignWcsStep’, ‘FlatFieldStep’ and ‘SourceTypeStep’; no photometric calibration was applied. The next steps were made using our own routines. We computed the background using two rectangular regions, one on each side of the spectral trace, between pixels 13 and 27 and between pixels 53 and 72 in the spatial direction, respectively. We computed the background value for each row (rows are along the spatial direction) in each region using a biweight location, averaged the two values and subtracted it from the full row. This background subtraction was done for each integration. Then, we extracted the stellar flux using aperture photometry by averaging pixels between 33 and 42 in each row to obtain the stellar spectrum at each integration. We also averaged pixels between 33 and 42 in the spatial direction and between 5 μm and 10.5 μm in the spectral direction to obtain the broadband flux. We averaged the spectra in 11 0.5-μm-wide wavelength channels. For each channel and for the broadband light curve, we normalized the light curve using the second eclipse as a reference flux, computed a running median filter using a 100-point window size, and rejected points that were more than 3 σ away from that median using a 5-iteration sigma-clipping. To limit the impact of the initial ramp on the fitting, we trim the first 779 integrations from the broadband light curve and a similar number of integrations for each channel (the exact number depends on the channel). Finally, we subtracted 1 from the normalized light curves to have the secondary eclipse flux centred on 0. These cleaned light curves were used for phase curve, eclipse and transit fits.

SPARTA reduction

We reduced the data with the open-source Simple Planetary Atmosphere Reduction Tool for Anyone (SPARTA), first introduced in ref. 48 to analyse the MIRI LRS phase curve of GJ 1214b. We started from the uncalibrated data and proceeded all the way to the final results without using any code from the jwst or Eureka! pipelines. In stage 1, we started by discarding the first five groups as well as the last group, because these groups show anomalies due to the reset switch charge decay and the last-frame effect. We fitted a slope to the up-the-ramp reads in every pixel of every integration in every exposure. We calculated the residuals of these linear fits, and for every pixel, we computed a median residual for every group across all integrations. This ‘median residual’ array has dimensions N grp × N rows × N cols . This array was subtracted from the original uncalibrated data and the up-the-ramp fit was redone, this time without discarding any groups except those that were more than 5 σ away from the best-fit line. Such outliers, which may be due to cosmic rays, were discarded and the fit recomputed until convergence. This procedure straightens out any nonlinearity in the up-the-ramp reads that is consistent across integrations, such as the reset switch charge decay, the last-frame effect or inaccuracies in the nonlinearity coefficients. After up-the-ramp fitting, we removed the background by removing the mean of columns 10–24 and 47–61 (inclusive, zero-indexed) for every row of every integration. As these two regions are of equal size and equally distant from the trace, any linear spatial trend in the background is naturally removed.

In the next step, we computed a pixel-wise median image over all integrations. This median image was used as a template to determine the position of the trace in each integration, by shifting and scaling the template until it matched the integration (and minimizes the χ 2 ). It was also used as the point spread profile for optimal extraction, after shifting in the spatial direction by the amount calculated in the previous step. Outliers more than 5 σ discrepant from the model image (which may be cosmic rays) were masked, and the optimal extraction was repeated until convergence. The z -scores image (image minus model image all divided by expected error, including photon noise and read noise) have a typical standard deviation of 0.88, compared with a theoretical minimum value of 1, indicating that the errors are being overestimated.

After optimal extraction, we gathered all the spectra and positions into one file. To reject outliers, we created a broadband light curve, detrended it by subtracting a median filter with a width 100 times less than the total data length and rejected integrations greater than 4 σ away from 0 (which may be cosmic rays). Sometimes only certain wavelengths of an integration are bad, not the entire integration. We repaired these by detrending the light curve at each wavelength, identifying 4 σ outliers and replacing them with the average of their two immediate temporal neighbours.

Spectral binning

To investigate the effects of spectral binning, we utilized the MIRI time-series commissioning observations of the transit of L168-9b (JWST-COM/MIRI-1033; ref. 29 ). L168-9b was chosen to have a clear transit signal while also having no detectable atmospheric signatures expected in its mid-infrared transmission spectrum; as a result, the observed scatter in the transmission spectrum can be used as an independent measurement of the uncertainties in the transit depths. The same procedure cannot be done on our WASP-43b science observations as there may be detectable atmospheric signatures.

Following the Eureka! reduction methods described by ref. 29 , we tried binning the L168-9b spectroscopic light curves at different resolutions and compared the observed standard deviation of the transmission spectrum with the median of the transit depth uncertainties estimated from fitting the spectral light curves. As shown in Extended Data Fig. 1 , the uncertainties in the pixel-level light curves underestimate the scatter in the transmission spectrum by a factor of about two. Because pairs of rows (in MIRI’s rotated reference frame) are reset together, it is reasonable to assume that there could be odd–even effects that would average out if combining pairs of rows; indeed, there do appear to be differences in the amplitude of the initial exponential ramp feature between odd and even rows. However, combining pairs of rows still leads to appreciable underestimation of the scatter in the transmission spectrum. Interestingly, the underestimation of the uncertainties appears to decrease with decreasing wavelength resolution. This is likely explained by wavelength-correlated noise that gets averaged out with coarse binning. A likely culprit for this wavelength-correlated noise may be the 390 Hz periodic noise observed in several MIRI subarrays, which causes clearly structured noise with a period of ~9 rows 29 (M. Ressler, private communication); this noise source is believed to be caused by MIRI’s electronics and possible mitigation strategies are still under investigation. Until the source of the excess wavelength-correlated noise is definitively determined and a noise mitigation method is developed, we recommend that MIRI/LRS observations should be binned to a fairly coarse spectral resolution as this gives better estimates of the uncertainties and also gives spectra that are closer to the photon-limited noise regime. However, we caution against quantitative extrapolations of the uncertainty underestimation to other datasets; because we do not know the source of the excess noise, we do not know how it might change with different parameters such as groups per integration or stellar magnitude.

Ultimately, for each reduction method, we binned the spectra down to a constant 0.50-μm-wavelength grid spanning 5–12 μm, giving a total of 14 spectral channels. However, as is described below, we only end up using the 11 spectral channels spanning 5–10.5 μm for science. This 0.5-μm-binning scheme combines 7 wavelengths for the shortest bin and 25 wavelengths for the longest bin, which has the added benefit of binning down the noise at longer wavelengths where there are fewer photons. However, even for this coarse of a binning scheme, we do expect there to be some additional noise beyond our estimated uncertainties on the spectrum of WASP-43b (Extended Data Fig. 1 ). As the structure of this noise source is not well understood nor is the extent to which our error bars are underestimated, our best course of action was to consider error inflation when performing spectroscopic inferences (described in more detail below).

Light-curve fitting

Detrending the initial exponential ramp.

As with other MIRI/LRS observations 29 , our spectroscopic light curves showed a strong exponential ramp at the start of the observations. As shown in Extended Data Fig. 2 , the strength and sign of the ramp varies with wavelength, changing from a strong downwards ramp at 5 μm to a nearly flat trend around 8 μm, and then becoming an upwards ramp towards longer wavelengths. From 10.6 μm to 11.8 μm, the ramp timescale became much longer and the amplitude of the ramp became much stronger; this region of the detector was previously discussed 44 and is discussed in more detail below. In general, most of the ramp’s strength had decayed within ~30–60 min, but at the precision of our data, the residual ramp signature still had an important impact on our nightside flux measurements due to the similarity of the ramp timescale with the orbital period. Unlike in the case of the MIRI/LRS commissioning observations of L168-9b 29 , we were not able to safely fit the entire dataset with a small number of exponential ramps. When fitting the entire dataset, we found that non-trivial choices about the priors for the ramp amplitudes and timescales resulted in significantly different spectra at phases 0.75 (morning hemisphere) and especially 0.0 (nightside); because the dayside spectrum is measured again near the end of the observations, it was less affected by this systematic noise.

Ultimately, we decided to conservatively discard the first 779 integrations (134.2 min), leaving only one transit duration of baseline before the first eclipse ingress began. After removing the initial 779 integrations, we found that a single exponential ramp model with broad priors that varied freely with wavelength was adequate to remove the signature. In particular, after removing the first 779 integrations we found that our dayside and nightside emission spectra were not significantly affected by (1) fitting two exponential ramps instead of one, (2) adjusting our priors on the ramp timescale to exclude rapidly decaying ramps with timescales >15 d −1 instead of >100 d −1 , (3) putting a uniform prior on the inverse timescale instead of the timescale, or (4) altering the functional form of the ramp by fitting for an exponential to which the time was raised. After removing the first ~2 h, we also found that the ramp amplitude and timescale did not vary strongly with wavelength (excluding the ‘shadowed region’ described below), although fixing these parameters to those fitted to the broadband light curve affected several points in the nightside spectrum by more than 1 σ ; we ultimately decided to leave the timescale and amplitude to vary freely with wavelength as there is no a priori reason to assume that they should be equal. With careful crafting of priors, it appeared possible to get results similar to our final spectra while removing only the first few integrations, but trimming more integrations and only using a single exponential ramp model required fewer carefully tuned prior assumptions for which we have little physical motivation.

Shadowed region effect

As was described in ref. 44 , we also identified a strong discontinuity in the spectroscopic light curves spanning pixel rows 156–220 (10.6–11.8 μm) in these observations. In this range, the temporal behaviour of the detector abruptly changes to a large-amplitude, long-timescale, upwards ramp that appears to slightly overshoot before decaying back down and approaching an equilibrium. These pixels coincide with a region of the detector between the Lyot coronagraph region and the four-quadrant phase mask region, which is unilluminated except when the dispersive element is in the optical path; as a result, we have taken to calling this unusual behaviour as the ‘shadowed region effect’. Strangely, not all MIRI/LRS observations show this behaviour, with the MIRI/LRS commissioning time-series observations 29 and the GJ 1214b phase-curve observations 48 showing no such effect. In fact, we know of only two other observations that show similar behaviour: the observation of the transit of WASP-80b (JWST-GTO-1177; T. Bell, private communication) and the observation of the phase curve of GJ 367b (JWST-GO-2508; M. Zhang, private communication). Informatively, the eclipse observation of WASP-80b taken 36 h after the WASP-80b transit using the same observing procedure (JWST-GTO-1177; T. Bell, private communication) did not show the same shadowed region effect, indicating that the effect is unlikely to be caused by stray light from nearby stars or any other factors that stayed the same between those two observations. Our best guess at this point is that the effect is related to the illumination history of the detector and the filter used by the previous MIRI observation (because the detector is illuminated at all times, even when it is not in use), but this is still under investigation and at present there is no way of predicting whether or not an observation will be impacted by the shadowed region effect. It is important to note, however, that from our limited knowledge at present that the shadowed region effect appears to be either present or not, with observations either strongly affected or seemingly completely unaffected.

Using the general methods described in the Eureka! v1 fit, we attempted to model the shadowed region effect with a combination of different ramp models, but nothing we tried was able to cleanly separate the effect from the phase variations, and there was always some clear structure left behind in the residuals of the fit. Another diagnostic that our detrending attempts were unsuccessful was that the phase offset as a function of wavelength smoothly varied around ~10° E in the unaffected region of the detector, but in the shadowed region, the phase offset would abruptly change to ~5° W; such a sharp change in a suspect region of the detector seems highly unlikely to be astrophysical in nature. As a result, we ultimately chose to exclude the three spectral bins spanning 10.5–12 μm from our retrieval efforts.

Determining the number of sinusoid harmonics

To determine the complexity of the phase-curve model required to fit the data, we used the Eureka! v1 reduction and most of the Eureka! v1 fitting methods described below, with the exception of using the dynesty 49 nested sampling algorithm (which computes the Bayesian evidence, ${{{\mathcal{Z}}}}$ ) and a batman transit and eclipse model. Within dynesty, we used 256 live points, ‘multi’ bounds, ‘rwalk’ sampling, and ran until the estimated $d\ln ({{{\mathcal{Z}}}})$ reached 0.1. We then evaluated first-, second- and fourth-order models for the broadband light curve, excluding all third-order sinusoidal terms from the fourth-order model as these terms are not likely to be produced by the planet’s thermal radiation 50 , 51 . We then compared the Bayesian evidences of the different models following refs. 52 , 53 and found that the second-order model was significantly preferred over the first-order model at 16 σ ( ${{\Delta }}\ln ({{{\mathcal{Z}}}})=128$ ), while the second-order model was insignificantly preferred over the fourth-order model at 2.2 σ ( ${{\Delta }}\ln ({{{\mathcal{Z}}}})=1.3$ ). This is also confirmed by eye where the first-order model leaves clear phase-variation signatures in the residuals, while the residuals from the second-order model leave no noticeable phase variations behind. Finally, we also compared the phase-resolved spectra obtained from different order phase-curve models; we found that our spectra significantly changed going from a first- to second-order model (altering one or more spectral points by >1 σ ), but the fourth-order model did not significantly change the resulting phase-resolved spectra compared with the second-order. As a result, the final fits from all reductions used a second-order model. The broadband light curves obtained from the four reductions and the associated phase-curve models are shown in Supplementary Fig. 1 .

Eureka! v1 fitting methods

We first sigma-clipped any data points that were 4 σ discrepant from a smoothed version of the data (made using a boxcar filter with a width of 20 integrations) to remove any obviously errant data points while preserving the astrophysical signals like the transit.

Our astrophysical model consisted of a starry 54 transit and eclipse model, as well as a second-order sinusoidal phase-variation model. The complete astrophysical model had the form

where t is the time, F * is the received stellar flux (and includes the starry transit model), F day is the planetary flux at mid-eclipse, E ( t ) is the starry eclipse model (neglecting eclipse mapping signals for the purposes of this paper), and Ψ ( ϕ ) is the phase-variation model of the form

where ϕ is the orbital phase in radians with respect to eclipse, and AmpCos1, AmpSin1, AmpCos2 and AmpSin2 are all fitted coefficients. The second-order phase-variation terms allow for thermal variations across the face of the planet that are more gradual or steep than a simple first-order sinusoid would allow. We numerically computed dayside, morning, nightside and evening spectra using the above Ψ ( ϕ ) function at ϕ = 0, π/2, π and 3π/2, respectively. To allow the starry eclipse function to account for light travel time, we used a stellar radius ( R * ) of 0.667 R ⊙ (ref. 55 ) to convert the fitted a / R * (the scaled semi-major axis) to physical units. For our transit model, we used a reparameterized version of the quadratic limb-darkening model 56 with coefficients u 1 and u 2 uniformly constrained between 0 and 1, and used a minimally informative prior on the planet-to-star radius ratio ( R p / R * ).

Our systematics model consisted of a single exponential ramp in time to account for the idle-recovery drift documented for MIRI/LRS time-series observations 29 , a linear trend in time, and a linear trend with the spatial position and PSF width. The full systematics model can be written as

The linear trend in time is modelled as

where t l is the time with respect to the mid-point of the observations and where c 0 and c 1 are coefficients. The exponential ramp is modelled as

where t r is the time with respect to the first integration and where r 0 and r 1 are coefficients. The linear trends as a function of spatial position, y , are PSF width s y are modelled as

where f and g are coefficients. The linear trends as a function of spatial position and PSF width are with respect to the mean-subtracted spatial position and PSF width. Finally, we also fitted a multiplier (scatter mult ) to the estimated Poisson noise level for each integration to allow us to account for any noise above the photon limit as well as an incorrect value for the gain applied in stage 3.

With an initial fit to the broadband light curve (5–10.5 μm), we assumed a zero eccentricity and placed a Gaussian prior on the planet’s orbital parameters (period, P ; linear ephemeris, t 0 ; inclination, i ; and scaled semi-major axis, a / R * ) based on previously published values for the planet 30 . For the fits to the spectroscopic phase curves, we then fixed these orbital parameters to the estimated best fit from the broadband light curve fit to avoid variations in these wavelength-independent values causing spurious features in the final spectra. We fitted the observations using the No U-Turns Sampler (NUTS) from PyMC3 57 with 3 chains, each taking 6,000 tuning steps and 6,000 production draws with a target acceptance rate of 0.95. We used the Gelman–Rubin statistic 58 to ensure the chains had converged. We then used the 16th, 50th and 84th percentiles from the PyMC3 samples to estimate the best-fit values and their uncertainties.

Eureka! v2 fitting methods

For the second fit made with Eureka!, we proceeded very similarly to the Eureka! v1 fit. We clipped the light curves using a boxcar filter of 20 integrations wide with a maximum of 20 iterations and a rejection threshold of 4 σ to reject these outliers. We also modelled the phase curve using a second-order sinusoidal function, but we modelled the transit and eclipse using batman 59 instead of starry. Like in the Eureka! v1 fit, we modelled instrumental systematics with a linear polynomial model in time (equation ( 4 )), an exponential ramp (equation ( 5 )), a first-order polynomial in y position (equation ( 6 )) and a first-order polynomial in PSF width in the s y direction (equation ( 7 )).

We fitted the data using the emcee sampler 60 instead of NUTS, with 500 walkers and 1,500 steps. The jump parameters that we used were the same as in the Eureka! v1 fit: R p / R * , F day , u 1 , u 2 , AmpCos1, AmpSin1, AmpCos2, AmpSin2, c 0 , c 1 , r 0 , r 1 , f , g and scatter mult (multiplier to the estimated Poisson noise level for each integration like in the Eureka! v1 fit). We used uniform priors on u 1 and u 2 from 0 to 1, uniform priors on AmpCos1, AmpSin1, AmpCos2, AmpSin2 from −1.5 to 1.5 and broad normal priors and all other jump parameters. Convergence, mean values and uncertainties were computed in the same way as for the Eureka! v1 fit.

TEATRO fitting methods

To measure the planet’s emission as a function of longitude, we modelled the light curves with a phase-variation model, an eclipse model, a transit model and an instrument systematics model. The phase-curve model, Ψ ( t ), consists of two sinusoids: one at the planet’s orbital period, P , and one at P /2 to account for second-order variations. The eclipse model, E ( t ), and transit model, T ( t ), are computed with the exoplanet 61 package that uses the starry package 54 . We save the eclipse depth, δ e , and normalize E ( t ) to a value of 0 during the eclipse and 1 out of the eclipse, which we then call E N ( t ). We used published transit ephemerides 62 , a null eccentricity and published stellar parameters 63 . The planet-to-star radius ratio, R p / R * , impact parameter, b , and mid-transit time, t 0 , are obtained from a fit to the broadband light curve. The systematics model, S ( t ), is composed of a linear function to account for a downwards trend and an exponential function to account for the initial ramp. The full model is expressed as:

where Ψ ( t e ) is the value of Ψ at the mid-eclipse time, t e .

We fit our model to the data using a Markov chain Monte Carlo (MCMC) procedure based on the PyMC3 package 57 and gradient-based inference methods as implemented in the exoplanet package 61 . We set normal priors for t 0 , R p / R * , the stellar density ( ρ * ), a Ψ , b Ψ , c S and d S with mean values obtained from an initial nonlinear least-squares fit, a normal prior for a S with a zero mean, uniform priors for the surface brightness ratio between the planet’s dayside and the star ( s ), b , c Ψ and d Ψ , uninformative priors for the quadratic limb-darkening parameters 56 , and allowed for wide search ranges. We ran two MCMC chains with 5,000 tuning steps and 100,000 posterior samples. Convergence was obtained for all parameters (except in one case where a S was negligible and b S was unconstrained). We merged the posterior distributions of both chains and used their median and standard deviation to infer final values and uncertainties for the parameters. We also verified that the values obtained from each chain were consistent.

For the spectroscopic light-curve fits, we fixed all physical parameters to those obtained from the broadband light-curve fit except the surface brightness ratio, s , that sets the eclipse depth, we masked the transit part of the light curve, and used a similar procedure. After the fits, we calculated the eclipse depth, δ e , as s × ( R p / R * ) 2 , and computed Ψ ( t ) for the final parameters, Ψ f ( t ). The planetary flux is Ψ f ( t ) − Ψ f ( t e ) + δ e . We computed the uncertainty on the eclipse depth in two different ways: from the standard deviation of the posterior distribution of s × ( R p / R * ) 2 , and from the standard deviation of the in-eclipse points divided by $\sqrt{{N}_{{\mathrm{e}}}}$ , where N e is the number of in-eclipse points, and took the maximum of the two. To estimate the uncertainty on the planet’s flux, we computed the 1 σ interval of Ψ ( t ) based on the posterior distributions of its parameters, computed the 1 σ uncertainty of d S , and added them in quadrature to the uncertainty on the eclipse depth to obtain more conservative uncertainties.

The spectra presented in this paper and used in the combined spectra are based on system parameters that were derived from a broadband light curve obtained in the 5–12 μm range, a transit fit in which the stellar mass and radius were fixed, a simpler additive model in which the phase curve was not turned off during the eclipse, and an MCMC run that consisted in two chains of 10,000 tuning steps and 10,000 posterior draws. Updated spectra based on system parameters derived from the broadband light curve obtained in the 5–10.5 μm range, a transit fit that has the stellar density as a free parameter, the light-curve model shown in equation ( 8 ), and two MCMC chains of 5,000 tuning steps and 100,000 posterior draws are consistent within 1 σ at every point with those shown here. As we average four reductions and inflate the uncertainties during the retrievals, the impact of these updates on our results are expected to be marginal.

SPARTA fitting methods

We use emcee 60 to fit a broadband light curve with the transit time, eclipse time, eclipse depth, four phase-curve parameters ( C 1 and D 1 for the first-order, and C 2 and D 2 for the second-order sinusoids), transit depth, a / R * , b , flux normalization, error-inflation factor, instrumental ramp amplitude ( A ) and 1/timescale ( τ ), linear slope in time ( m ) with respect to the mean of the integration times $(\overline{t})$ , and linear slope with trace y position ( c y ) as free parameters. The best-fit transit and eclipse times, a / R * and b are fixed for the spectroscopic light curves.

For the spectroscopic fits, we then use emcee to fit the free parameters: the eclipse depth, four phase-curve parameters, error-inflation factor, flux normalization, instrumental ramp amplitude and 1/timescale, linear slope with time, and linear slope with trace y position. All parameters are given uniform priors. 1/timescale is given a prior of 5–100 d −1 , but the other priors are unconstraining. In summary, the instrumental model is:

while the planetary flux model is:

where E is the eclipse depth and ω = 2π/ P is the planet’s orbital angular frequency. Note that the phase variations were set to be zero during eclipse.

Combining independent spectra

Comparing the phase-resolved spectra from each reduction (Supplementary Fig. 2 ), we see that for wavelengths below 10.5 μm, the spectra are typically consistent, while larger differences arise in the >10.5 μm region affected by the shadowed region effect. For our final, fiducial spectrum, we decided to use the median spectrum and inflated our uncertainties to account for disagreements between different reductions. The median phase-resolved spectra were computed by taking the median F p / F * per wavelength. The uncertainties were computed by taking the median uncertainty per wavelength, and then adding in quadrature the RMS between the individual reductions and the median spectrum; this minimally affects the uncertainties where there is minimal disagreement and appreciably increases the uncertainties where the larger disagreements arise.

Each reduction also computed a transmission spectrum (Supplementary Fig. 2 ), which appears quite flat (within uncertainties) with minimal differences between reductions. WASP-43b is not an excellent target for transmission spectroscopy, however, and these transmission spectra are not expected to be overly constraining.

Atmospheric forward models

GCMs were used to simulate atmospheric conditions, from which synthetic phase curves and emission spectra were forward modelled and compared with the observations. The GCMs used in this study are listed in Supplementary Table 1 , and details of each simulation are provided in Extended Data Table 1 and the following sections.

Generic PCM

The Generic Planetary Climate Model (Generic PCM) is a three-dimensional global climate model designed for modelling the atmosphere of exoplanets and for palaeoclimatic studies. The model has been used for the study of planetary atmospheres of the Solar System 64 , 65 , 66 , terrestrial exoplanets 67 , mini-Neptunes 68 and hot Jupiters 69 . The dynamical core solves the primitive equations of meteorology on a Arakawa C grid. The horizontal resolution is 64 × 48 (that is, 5.625 × 3.75°) with 40 vertical levels, equally spaced in logarithmic scale between 10 Pa and 800 bars. Along with the various parameterizations of physical processes described in refs. 64 , 65 , 66 , 67 , 68 , the Generic PCM treats clouds as radiatively active tracers of fixed radii.

The model is initialized using temperature profiles from the radiative–convective one-dimensional model Exo-REM 70 . The radiative data are computed offline using the out-of-equilibrium chemical profiles of the Exo-REM run. We use 27 frequency bins in the stellar channel (0.261–10.4 μm) and 26 in the planetary channel (0.625–324 μm), all bins including 16 k -coefficients. We start the model from a rest state (no winds), with a horizontally homogeneous temperature profile. Models are integrated for 2,000 days, which is long enough to complete the spin-up phase of the simulation above the photosphere. We do not include Rayleigh drag in our models. The simulations are performed including clouds of Mg 2 SiO 4 , with varying cloud radii (0.1, 0.5, 1, 3 and 5 μm). We also computed cloudless and Mg 2 SiO 4 models with a 10× solar metallicity and the same radii for the cloud particles. Regardless of the composition and size of the clouds, our model clearly indicates that there is no cloud formation on the dayside. Asymmetric limbs are a natural result of our model, with the eastern terminator being warmer while the western limb is cloudier and cooler. Spectral phase curves were produced using the Pytmosph3R code 71 .

SPARC/MITgcm with radiative transfer post-processing by gCMCRT

SPARC/MITgcm couples a state-of-the-art non-grey, radiative-transfer code with the MITgcm 33 . The MITgcm solves the primitive equations of dynamical meteorology on a cubed-sphere grid 72 . It is coupled to the non-grey radiative-transfer scheme based on the plane-parallel radiative-transfer code of ref. 73 . The stellar irradiation incident on WASP-43b is computed with a PHOENIX model 74 , 75 , 76 . We use previously published opacities 77 , including more recent updates 78 , 79 , and the molecular abundances are calculated assuming local chemical equilibrium 80 . In the GCM simulations, the radiative-transfer calculations are performed on 11 frequency bins ranging from 0.26 μm to 300 μm, with 8 k -coefficients per bin statistically representing the complex line-by-line opacities 3 . The strong visible absorbers TiO and VO are excluded in our k tables similar to our previous GCMs of WASP-43b 3 , 23 that best match the observed dayside emission spectrum and photometry.

Clouds in the GCM are modelled as tracers that are advected by the flow 81 and can settle under gravity. Their formation and evaporation are subjected to chemical equilibrium predictions, that is, the condensation curves of various minerals described in ref. 80 . The conversion between the condensable ‘vapour’ and clouds is treated as a simple linear relaxation over a short relaxation timescale of 100 s. The scattering and absorption of the spatial- and time-dependent clouds are included in both the thermal and visible wavelengths of the radiative transfer. A similar dynamics–cloud–radiative coupling has been developed in our previous GCMs with simplified radiative transfer and has been used to study the atmospheric dynamics of brown dwarfs 9 , 82 and ultrahot Jupiters 83 . Clouds are assumed to follow a log-normal size distribution 84 , which is described by the reference radius r 0 and a non-dimensional deviation σ : $n(r)=\frac{{{{\mathcal{N}}}}}{\sqrt{2\uppi }\sigma r}\exp \left(-\frac{{[\ln (r/{r}_{0})]}^{2}}{2{\sigma }^{2}}\right)$ , where n ( r ) is the number density per radius bin of r and ${{{\mathcal{N}}}}$ is the total number density. σ and r 0 are free parameters and the local ${{{\mathcal{N}}}}$ is obtained from the local mass mixing ratio of clouds. The size distribution is held fixed throughout the model and is the same for all types of cloud.

Our GCMs do not explicitly impose a uniform radiative heat flux at the bottom boundary but rather relax the temperature of the lowest model layer (that is, the highest pressure layer) to a certain value over a short timescale of 100 s. This assumes that the deep GCM layer reaches the convective zone and the temperature there is set by the interior convection that ties to the interior structure of the planet. This lowest-layer temperature is in principle informed by internal structure models of WASP-43b, which are run by MESA hot Jupiter evolution modules 12 to match the present radius of WASP-43b. In most models, this lowest-layer temperature is about 2,509 K at about 510 bars. The horizontal resolution of our GCMs is typically C48, equivalent to about 1.88° per grid cell. The vertical domain is from 2 × 10 −4 bar at the top to 700 bars at the bottom and is discretized to 53 vertical layers. We typically run the simulation for over 1,200 days and average all physical quantities over the last 100 days of the simulations.

All our GCMs assume a solar composition. We performed a baseline cloudless model and one case with only MnS and Na 2 S clouds with r 0 = 3 μm, and then a few cases with MnS, Na 2 S and MgSiO 3 clouds with r 0 = 1, 1.5, 2 and 3 μm. The σ is held fixed at 0.5 in all our cloudy GCMs.

We post-process our GCM simulations with the state-of-the-art gCMCRT code, which is a publicly available hybrid Monte Carlo radiative transfer (MCRT) and ray-tracing radiative-transfer code. The model is described in detail in ref. 85 and has been applied to a range of exoplanet atmospheres 83 , 86 . gCMCRT can natively compute albedo, transmission and emission spectra at both low and high spectral resolution. gCMCRT uses custom k tables, which take cross-section data from both HELIOS-K 87 and EXOPLINES 88 . Here, we apply gCMCRT to compute low-resolution emission spectra and phase curves at R ≈ 300 from our GCM simulations. We use the three-dimensional temperature and condensate cloud tracer mixing ratio from the time-averaged end-state of each case. We assume the same cloud particle size distribution as our GCMs.

expeRT/MITgcm

The GCM expeRT/MITgcm uses the same dynamical core as SPARC/MITgcm and solves the hydrostatic primitive equations on a C32 cubed-sphere grid 72 . It resolves the atmosphere above 100 bar on 41 log-spaced cells between 1 × 10 −5 bar and 100 bar. Below 100 bar, 6 linearly spaced grid cells between 100 bar and 700 bar are added. The model expeRT/MITgcm thus resolves deep dynamics in non-inflated hot Jupiters like WASP-43b 16 , 89 .

The GCM is coupled to a non-grey radiative-transfer scheme based on petitRADTRANS 90 . Fluxes are recalculated every fourth dynamical time step. Stellar irradiation is described by the spectral fluxes from the PHOENIX model atmosphere suite 74 , 75 , 76 . The GCM operates on a precalculated grid of correlated k -binned opacities. Opacities from the ExoMol database 91 are precalculated offline on a grid of 1,000 logarithmically spaced temperature points between 100 K and 4,000 K for every vertical layer. We further include the same species as shown in ref. 89 except TiO and VO to avoid the formation of a temperature inversion in the upper atmosphere. These are: H 2 O (ref. 92 ), CH 4 (ref. 93 ), CO 2 (ref. 94 ), NH 3 (ref. 95 ), CO (ref. 96 ), H 2 S (ref. 97 ), HCN (ref. 98 ), PH 3 (ref. 99 ), FeH (ref. 100 ), Na (refs. 74 , 101 ) and K (refs. 74 , 101 ). For Rayleigh scattering, the opacities are H 2 (ref. 102 ) and He (ref. 103 ), and we add the following collision-induced absorption (CIA) opacities: H 2 –H 2 (ref. 104 ) and H 2 –He (ref. 104 ). We use for radiative-transfer calculations in the GCM the same wavelength resolution as SPARC/MITgcm (S1), but incorporate 16 instead of 8 k -coefficients. Two cloud-free WASP-43b GCM simulations were performed, one with solar and one with 10× solar element abundances. Each simulation ran for 1,500 days to ensure that the deep wind jet has fully developed. The GCM results used in this paper were time averaged over the last 100 simulation days.

Spectra and phase curves are produced from our GCM results in post-processing with petitRADTRANS 90 and prt_phasecurve 89 using a spectral resolution of R = 100 for both the phase curve and the spectra.

Originally adapted from the GCM of ref. 105 by refs. 106 , 107 , 108 , the RM-GCM model has been applied to numerous investigations of hot Jupiters and mini-Neptunes 35 , 109 , 110 , 111 . The GCM’s dynamical core solves the primitive equations of meteorology using a spectral representation of the domain, and it is coupled to a two-stream, double-grey radiative-transfer scheme based on ref. 112 . Recent updates have added aerosol scattering 35 with radiative feedback 8 , 36 . Following ref. 8 , aerosols are representative of condensate clouds and are treated as purely temperature-dependent sources of opacity, with constant mixing ratios set by the assumed solar elemental abundances. The optical thicknesses of the clouds are determined by converting the relative molecular abundances (or partial pressures) of each species into particles with prescribed densities and radii 8 . The model includes up to 13 different cloud species of various condensation temperatures, abundances and scattering properties. Places where clouds overlap have mixed properties, weighted by the optical thickness of each species.

Simulations from this GCM included a clear atmosphere and two sets of cloudy simulations. Following ref. 8 , one set of cases included 13 different species: KCl, ZnS, Na 2 S, MnS, Cr 2 O 3 , SiO 2 , Mg 2 SiO 4 , VO, Ni, Fe, Ca 2 SiO 4 , CaTiO 2 and Al 2 O 3 ; the other set omitted ZnS, Na 2 S, MnS, Fe and Ni, based on considerations of nucleation efficiency 113 . For both cloud composition scenarios, the models explored the observational consequences of variations in the cloud deck’s vertical thickness through a series of simulations with clouds tops truncated over a range of heights at 5-layer intervals (roughly a scale height), ranging from 5 to 45 layers of the 50-layer model. This effectively mimics a range of vertical mixing strengths. From the complete set published in ref. 26 , we selected a subset, with clouds of maximum vertical extents between two and nine scale heights from each of the two cloud composition scenarios.

Simulations were initialized with clear skies, no winds and no horizontal temperature gradients. We ran the simulations for over 3,500 planetary orbits, assuming tidal synchronization. Resulting temperature, wind and cloud fields of the GCM were then post-processed 114 , 115 to yield corresponding emission phase curves.

THOR 116 , 117 is an open-source GCM developed to study the atmospheres and climates of exoplanets, free from Earth- or Solar System-centric tunings. The core that solves the fluid flow equations, the dynamical core, solves the non-hydrostatic compressible Euler equations on an icosahedral grid 116 , 118 . THOR has been validated and used to simulate the atmosphere of Earth 116 , 119 , Solar System planets 120 , 121 and exoplanets 116 , 117 , 122 .

For this work, THOR used the same configuration as with previously published simulations to study the atmospheric temperature structure, cloud cover and chemistry of WASP-43b 4 , 38 , 123 . Two simulations were conducted, one with a clear atmosphere and another with a cloud structure on the nightside of the planet. To represent the radiative processes, THOR uses a simple two-band formulation calibrated to reproduce the results from more complex non-grey models on WASP-43b 3 , 124 . A simple cloud distribution on the nightside of the planet and optical cloud properties are parameterized 4 and adapted to reproduce previous HST 21 and Spitzer 4 , 22 observations. These simulations on WASP-43b with THOR have also been used to test the performance of future Ariel phase-curve observations 125 .

Both simulations, with clear and cloudy atmospheres, started with isothermal atmospheres (1,440 K, equilibrium temperature) and integrated for roughly 9,400 planetary orbits (assuming a tidally locked configuration) until a statistically steady state of the deep atmosphere thermal structure was reached. The long integration avoids biasing the results towards the set initial conditions 120 .

The multiwavelength spectra are obtained from post-processing the three-dimensional simulations with a multiwavelength radiative-transfer model 126 . The disk-averaged planet spectrum is calculated at each orbital phase by projecting the outgoing intensity for each geographical location of the observed hemisphere. The spectra include cross-sections of the main absorbers in the infrared, drawn from the ExoMOL (H 2 O (ref. 92 ), CH 4 (ref. 127 ), NH 3 (ref. 128 ), HCN (ref. 129 ) and H 2 S (ref. 97 )), HITEMP 130 (CO 2 and CO) and HITRAN 131 (C 2 H 2 ) databases. The Na and K resonance lines 132 are also added, as were H 2 –H 2 and H 2 –He CIA 104 . The atmospheric bulk composition was assumed to have solar abundance (consistent with HST/WFC3 spectrum observations), and each chemical species concentration was calculated with the FastChem model 133 . The PHOENIX models 74 , 75 , 76 were used for the WASP-43 star spectrum.

Atmospheric retrieval models

We perform atmospheric retrievals on the phase-resolved emission spectra using six different retrieval frameworks, each described in turn below. The chemical constraints from these analyses are summarized in Extended Data Tables 2 and 3 , and the spectral fits obtained are shown in Extended Data Fig. 3 . Across the six retrieval analyses, we use an error-inflation parameter to account for the effects of unknown data and/or model uncertainties. This free parameter is wavelength independent and multiplies the 1 σ error bars in the calculation of the likelihood function in the Bayesian sampling algorithm.

HyDRA retrieval framework

The HyDRA atmospheric retrieval framework 134 consists of a parametric atmospheric forward model coupled to PyMultiNest 135 , 136 , a nested sampling Bayesian parameter estimation algorithm 137 . HyDRA has been applied to hydrogen-rich atmospheres 138 , 139 , and further adapted for secondary atmospheres 140 and high-resolution spectroscopy in both one and two dimensions 141 , 142 . The input parameters for the atmospheric forward model include constant-with-depth abundances for each of the chemical species considered, six temperature profile parameters corresponding to the temperature profile model of ref. 143 , and a constant-with-wavelength multiplicative error-inflation parameter to account for model uncertainties. We additionally include a dilution parameter, A HS , for the dayside, morning and evening hemispheres, which multiplies the emission spectrum by a constant factor <1 and accounts for temperature inhomogeneities in each hemisphere 144 .

We consider opacity contributions from the chemical species that are expected to be present in hot Jupiter atmospheres and that have opacity in the MIRI LRS wavelength range: H 2 O (ref. 130 ), CH 4 (refs. 127 , 145 ), NH 3 (ref. 128 ), HCN (refs. 98 , 129 ), CO (ref. 130 ), CO 2 (ref. 130 ), C 2 H 2 (refs. 131 , 146 ), SO 2 (ref. 147 ), H 2 S (refs. 97 , 148 ) and CIA due to H 2 –H 2 and H 2 –He (ref. 104 ). The line-by-line absorption cross-sections for these species are calculated following the methods described in ref. 134 , using data from each of the references listed. We further explore retrievals with a simple silicate cloud model, which includes the modal particle size, cloud particle abundance, cloud base pressure and a pressure exponent for the drop-off of cloud particle number density with decreasing pressure. The opacity structure of the cloud is calculated using the absorption cross-sections of ref. 149 .

Given the input chemical abundances, temperature profile and cloud parameters, the forward model calculates line-by-line radiative transfer to produce the thermal emission spectrum at a resolution of R ≈ 15,000. The spectrum is then convolved to a resolution of 100, binned to the same wavelength bins as the observations and compared with the observed spectrum to calculate the likelihood of the model instance. The nested sampling algorithm explores the parameter space using 2,000 live points, and further calculates the Bayesian evidence of the retrieval model, which can be used to compare different models 52 . In particular, we calculate the detection significance of a particular chemical species by comparing retrievals that include/exclude that species, fixing the value of the error-inflation parameter to be the median retrieved value found with the full retrieval model.

Across the four phases, the only chemical species detected with statistical significance ( ≳ 3 σ ) is H 2 O. The retrieved H 2 O abundances are in the range ~30–10 4 ppm (1 σ uncertainties), with detection significances varying between ~3 σ and ~4 σ (Extended Data Table 2 ). We do not detect CH 4 at any phase, and place an upper limit of 16 ppm on the nightside CH 4 abundance, potentially indicating disequilibrium chemistry processes as described in the main text. We do not detect NH 3 at any phase either, consistent with the very low NH 3 abundances predicted by both chemical equilibrium and disequilibrium models 23 . The retrievals do not favour cloudy models over clear models with statistical significance, with extremely weak preferences of <1 σ at all phases. In addition, the posterior probability distributions for the cloud parameters are unconstrained. Extended Data Fig. 5 shows the pressure ranges of the atmospheric model probed by the observations.

PyratBay retrieval framework

PyratBay is an open-source framework that enables atmospheric modelling, spectral synthesis, and atmospheric retrievals of exoplanet observations 150 . The atmospheric model consists of parametric temperature, composition and altitude profiles as a function of pressure, for which emission and transmission spectra can be generated. The radiative-transfer module considers opacity from alkali lines 151 , Rayleigh scattering 152 , 153 , Exomol and HITEMP molecular line lists 130 , 154 pre-processed with the REPACK package 155 to extract the dominant line transitions, CIA 156 and cloud opacities. The PyratBay retrieval framework has the ability to stagger model complexity and explore a hierarchy of different model assumptions. Temperature models range from an isothermal profile to physically motivated parameterized models 143 , 157 . Composition profiles range from the simpler constant-with-altitude ‘free abundance’ to the more complex ‘chemically consistent’ retrievals, the latter accomplished via the TEA code 158 ; while cloud condensate prescriptions range from the classic ‘power law + grey’ to a ‘single-particle-size’ haze profile, a partial-coverage factor ‘patchy clouds’ 159 , and the complex parameterized Mie-scattering thermal stability cloud (TSC) model (J.B. et al., manuscript in preparation). The TSC cloud prescription, initially inspired by refs. 84 , 160 , has additional flexibility in the location of the cloud base and was further improved for this analysis (see below). The formulation utilizes a parameterized cloud shape, effective particle size and gas number density below the cloud deck, while the atmospheric mixing and settling are wrapped up inside the cloud extent and the condensate mole fraction as free parameters. This cloud model was applied to WASP-43b JWST/MIRI phase-curve simulations 23 , generated during the JWST preparatory phase, in anticipation of the actual WASP-43b JWST/MIRI observations. We showed that the TSC model has the ability to distinguish between MgSiO 3 and MnS clouds on the nightside of the planet.

For this analysis, we conducted a detailed investigation using various model assumptions. We started by exploring simple temperature prescriptions and gradually moved towards more complex ones. Initially, we considered opacity contributions from all chemical species expected to be observed in the MIRI wavelength range (H 2 O, CH 4 , NH 3 , HCN, CO, CO 2 , C 2 H 2 , SO 2 , H 2 S), but eventually focused on only those that are fit by the data. We also implemented the dilution parameter 144 and an error-inflation factor, which account for some additional model and data uncertainties. The constraints on H 2 O (together with the detection significance 161 ) and the upper limit for CH 4 for all phases are given in Extended Data Table 2 . The abundances of these species across all phases were largely model independent. However, the tentative constraints on NH 3 , which we saw in multiple phases, were strongly model dependent, and were completely erased with the inclusion of the dilution parameter and the error inflation, thus we do not report them here. WASP-43b emission spectra were computed at a resolution of R ≈ 15,000 utilizing opacity sampling of high-resolution pre-computed cross-sections ( R ≈ 10 6 ) of considered species. Furthermore, we thoroughly examined the possibility of detecting clouds in each of the four-quadrant phases, with a special emphasis on the nightside of the planet. To do this, we employed the TSC model, as in our previous analysis 23 , and explored a range of cloud species, MgSiO 3 , MnS, ZnS and KCl, that would condense under the temperature regimes expected for WASP-43b 162 (Extended Data Fig. 6 , left). We also introduced the effective standard deviation of the log-normal distribution 84 as a free parameter ( σ log ), allowing for even more flexibility in our cloud model (Extended Data Fig. 6 , right, last subpanel). To thoroughly explore the parameter space, we used two Bayesian samplers, the differential-evolution MCMC algorithm 163 , implemented following ref. 164 , and the nested sampling algorithm, implemented through PyMultiNest 135 , 136 , utilizing 15 million models and 2,000 live points, respectively. Our investigation did not provide constraints on any of the cloud parameters for any of the explored cloud condensates at any of the planetary phases, indicating the absence of detectable spectral features from clouds in the observations (Extended Data Fig. 6 , right).

NEMESIS retrieval framework

NEMESIS 165 , 166 is a free retrieval framework that uses a fast correlated- k 167 forward model, combined with either an optimal estimation or nested sampling retrieval algorithm. It has been used to perform retrievals on spectra of numerous planetary targets, both inside and outside the Solar System 168 , 169 . In this work, we use the PyMultiNest sampler 136 with 500 live points. The retrieval model presented includes four spectrally active gases, H 2 O (ref. 92 ), CO (ref. 96 ), CH 4 (ref. 93 ) and NH 3 (ref. 95 ), with k tables calculated as in ref. 91 ; we did not include CO 2 or H 2 S after initial tests indicated these were not required to fit the spectrum. All gases are assumed to be well mixed in altitude. CIA from H 2 and He is taken from refs. 156 , 170 . The spectrum is calculated at the resolution of the observation, using optimized channel integrated k tables generated from original k tables with a resolving power R = 1,000. The temperature profile is modelled as a three-parameter Guillot profile, after ref. 157 , with free parameters κ , γ and β ( α is fixed to be zero). We include a well-mixed, spectrally grey cloud with a scalable total optical depth with a cloud top at 12.5 mbar. The other retrieved parameters are a hotspot dilution factor for phases 0.25, 0.5 and 0.75, following ref. 144 , and an error-inflation term.

To calculate the detection significance for H 2 O, we run the retrieval with and without H 2 O, with all other aspects of the run identical. We then take the difference of the PyMultiNest global log-evidence values for the two scenarios, and convert from log(Bayesian evidence) to sigma following ref. 52 . The 99% upper limit for CH 4 is calculated from the equally weighted posterior distribution. We also attempt to retrieve CO and NH 3 abundances. CO is generally poorly constrained, and NH 3 is unconstrained for phases 0 and 0.75; for log(NH 3 ), we recover a 99% upper limit of −2.2 at phase 0.25 and −3.9 at phase 0.5. The cloud opacity is also generally unconstrained, with the total optical depth able to span several orders of magnitude. We stress that this model is very crude as it has only one variable cloud parameter, and further exploration of suitable cloud models for mid-infrared phase curves is warranted in future work.

SCARLET retrieval framework

We perform atmospheric retrievals on the four phase-resolved spectra using the SCARLET framework 160 , 171 . The planetary disk-integrated thermal emission, F p , is modelled for a given set of atomic/molecular abundances, temperature–pressure profile and cloud properties. We compare our model spectra with the observations by normalizing the thermal emission F p using a PHOENIX 74 , 75 , 76 stellar model spectrum with effective temperature T eff = 4,300 K and surface gravity log g = 4.50. The model spectra are computed at a resolving power of R = 15,625, convolved to the resolving power of MIRI/LRS and then binned to the 11 spectral bins (<10.5 μm) considered in the analysis, assuming the throughput to be uniform over a single bin.

The atmospheric analysis is performed considering thermochemical equilibrium, where the metallicity [M/H] ( ${{{\mathcal{U}}}}[-3,3]$ ) and carbon-to-oxygen ratio ( ${{{\mathcal{U}}}}[0,3]$ ) are free parameters that dictate the overall atmospheric composition. We use a free parameterization of the temperature–pressure profile 172 by fitting for N = 4 temperature points ( ${{{\mathcal{U}}}}[100,4400]\,{\mathrm{K}}$ ) with a constant spacing in log-pressure. The temperature–pressure profile is interpolated to the 50 layers ( P = 10 2 –10 −6 bar) considered in the model using a spline function to produce a smooth profile. We use a grid of chemical equilibrium abundances produced with FastChem2 173 to interpolate the abundance of species as a function of temperature and pressure for given values of [M/H] and C/O. The species considered in the equilibrium chemistry are H, H − (refs. 174 , 175 ), H 2 , He, H 2 O (ref. 92 ), OH (ref. 130 ), CH 4 (ref. 127 ), C 2 H 2 (ref. 176 ), CO (ref. 130 ), CO 2 (ref. 130 ), NH 3 (ref. 95 ), HCN (ref. 98 ), PH 3 (ref. 99 ), TiO (ref. 177 ) and VO (ref. 178 ). All opacities for these species are considered when computing the thermal emission. We account for potential spatial atmospheric inhomogeneities in the planetary disk that are observed at a given phase by including an area fraction parameter A HS ( ${{{\mathcal{U}}}}[0,1]$ ), which is meant to represent the possibility of a fraction of the disk contributing to most of the observed thermal emission 144 . This parameter is considered for all phases with the exception of the nightside, which is expected to be relatively uniform. Finally, we fit for an error-inflation parameter k σ ( ${{{\mathcal{U}}}}[0.1,10]$ ) to account for potential model and data uncertainty, which results in a total of 8 (7 for the nightside) free parameters. We consider 8 walkers per free parameter for the retrievals which are run for 30,000 steps. The first 18,000 steps are discarded when producing the posterior distributions of the free parameters.

PLATON retrieval framework

PLATON 179 , Planetary Atmosphere Tool for Observer Noobs, is a Bayesian retrieval tool that assumes equilibrium chemistry. We adopt the temperature–pressure profile parameterization of ref. 180 , and use the dynesty nested sampler 49 to retrieve the following free parameters: stellar radius; stellar temperature; the log metallicity, log( Z ); C/O; 5 temperature–pressure parameters (log( κ th ), log( γ ), log( γ 2 ), α , β ); and an error multiplier. The stellar radius and temperature are given Gaussian priors with means and standard deviations set by the measurements in ref. 55 : 4,400 ± 200 K and 0.667 ± 0.011 R ⊙ , respectively. The combination of the two have a similar effect to the dilution parameter of other retrieval codes, which multiplies the emission spectrum by a constant. For phase 0.0, we obtain a significantly better fit when methane opacity is set to zero (thus removing all spectral features from methane). We therefore adopt this as the fiducial model, whereas for other phases, we do not zero out any opacities.

For all retrievals, we use nested sampling with 1,000 live points. The opacities (computed at R = 10,000) and the line lists used to compute them are listed in ref. 179 . We include all 31 species in retrieval, notably including H 2 O, CO, CO 2 , CH 4 (except on the nightside), H 2 S and NH 3 .

ARCiS retrieval framework

ARCiS (Artful modelling code for exoplanet science) is an atmospheric modelling and Bayesian retrieval code 181 , 182 that utilizes the MULTINEST 135 Monte Carlo nested sampling algorithm. The code was used in previous retrievals of the atmosphere of WASP-43b in transmission 183 , using the observations of ref. 184 , and in phase-resolved emission 185 , using the observations of refs. 21 , 22 , 25 , 186 . Reference 183 found some evidence that AlO improves the fit of the transmission spectra of WASP-43b in the 1.1–1.6 μm region. We therefore include in our models for this work the following set of molecules in our free molecular retrievals: H 2 O (ref. 92 ), CO (ref. 96 ), CO 2 (ref. 94 ), NH 3 (ref. 95 ), CH 4 (ref. 93 ) and AlO (ref. 187 ). The molecular line lists are from the ExoMol 154 , 188 or HITEMP 130 databases as specified, and k tables from the ExoMolOP opacity database 91 . CIA for H 2 and He are taken from refs. 156 , 170 . We explore the inclusion of a variety of additional molecules that have available line list data with spectral features in the region of our observations, including HCN (ref. 98 ), SiO (ref. 189 ) and N 2 O (ref. 130 ). We use the Bayes factor, which is the difference between the nested sampling global log-evidence (log E ) between two models, to assess whether the inclusion of a particular parameter is statistically significant. For this, we run a retrieval with the base set of species only and another with the base set plus the molecule being assessed. The difference in log E between the two models is converted to a significance in terms of σ using the metric of ref. 52 . We explore the inclusion of a simple grey, patchy cloud model, which parameterizes cloud top pressure and degree of cloud coverage (from 0 for completely clear to 1 for completely covered). We use 1,000 live points and a sampling efficiency of 0.3 in MULTINEST for all retrievals.

We run retrievals both including and not including a retrieved error-inflation parameter. The error-inflation parameter is implemented as per ref. 190 to account for underestimated uncertainties and/or unknown missing forward model parameters. All phases apart from 0.0 retrieved a parameter that increases the observational error bars by two to three times their original values. The pressure–temperature profile parameterization of ref. 191 is used in all cases. We find evidence for the inclusion of H 2 O for all four phases, although this evidence goes from strong to weak when error inflation is included for the morning phase (0.75). We find no strong evidence for CH 4 at any phase, with 95% confidence upper limits on the log of the volume mixing ratio (VMR) of −4.9, −2.9, −3.2 and −2.2 for phases 0.0, 0.25, 0.5 and 0.75, respectively. We find some model-dependent hints of moderate evidence (based on the metric of ref. 52 ) of 4.4 σ for NH 3 at phase 0.5 (constrained to $\log{{\mathrm{VMR}}}={-4.5}_{-0.5}^{+0.7}$ ), 3.1 σ for CO at phase 0.5 ( $\log{{\mathrm{VMR}}}={-1.7}_{-0.7}^{+0.5}$ ) and 2.6 σ for CO at phase 0.25 ( $\log{{\mathrm{VMR}}}={-4.0}_{-0.4}^{+0.3}$ ). However, these disappear when the error-inflation parameter is introduced. We are not able to constrain any of the cloud parameters for any phase, and so do not find a statistical reason to include our simple cloud parameterization in the models to better fit the observations.

Data availability

The data used in this paper are associated with JWST DD-ERS programme 1366 (principal investigators N.M.B., J.L.B. and K.B.S.; observation 11) and are publicly available from the Mikulski Archive for Space Telescopes ( https://mast.stsci.edu ). Additional intermediate and final results from this work are archived on Zenodo at https://doi.org/10.5281/zenodo.10525170 (ref. 192 ).

Code availability

We used the following codes to process, extract, reduce and analyse the data: STScI’s JWST Calibration pipeline 46 , Eureka! 45 , TEATRO, SPARTA 48 , Generic PCM 64 , 65 , 66 , 67 , 68 , SPARC/MITgcm 9 , 33 , 81 , 82 , expeRT/GCM 16 , 72 , 89 , RM-GCM 8 , 35 , 106 , 107 , 108 , THOR 4 , 116 , 117 , 124 , 126 , 193 , HyDRA 134 , PyratBay 150 , NEMESIS 165 , 166 , SCARLET 160 , 171 , PLATON 179 , starry 54 , exoplanet 61 , PyMC3 57 , emcee 60 , dynesty 49 , numpy 194 , astropy 195 , 196 and matplotlib 197 .

Keating, D., Cowan, N. B. & Dang, L. Uniformly hot nightside temperatures on short-period gas giants. Nat. Astron. 3 , 1092–1098 (2019).

Article ADS Google Scholar

Beatty, T. G. et al. Spitzer phase curves of KELT-1b and the signatures of nightside clouds in thermal phase observations. Astron. J. 158 , 166 (2019).

Kataria, T. et al. The atmospheric circulation of the hot Jupiter WASP-43b: comparing three-dimensional models to spectrophotometric data. Astrophys. J. 801 , 86 (2015).

Mendonça, J. M., Malik, M., Demory, B.-O. & Heng, K. Revisiting the phase curves of WASP-43b: confronting re-analyzed Spitzer data with cloudy atmospheres. Astron. J. 155 , 150 (2018).

Parmentier, V. & Crossfield, I. J. M. in Handbook of Exoplanets (eds Deeg, H. J. & Belmonte, J. A.) 1419–1440 (Springer, 2018).

Hu, R., Demory, B.-O., Seager, S., Lewis, N. & Showman, A. P. A semi-analytical model of visible-wavelength phase curves of exoplanets and applications to Kepler-7b and Kepler-10b. Astrophys. J. 802 , 51 (2015).

Shporer, A. & Hu, R. Studying atmosphere-dominated hot Jupiter Kepler phase curves: evidence that inhomogeneous atmospheric reflection is common. Astron. J. 150 , 112 (2015).

Roman, M. T. et al. Clouds in three-dimensional models of hot Jupiters over a wide range of temperatures. I. Thermal structures and broadband phase-curve predictions. Astrophys. J. 908 , 101 (2021).

Tan, X. & Showman, A. P. Atmospheric circulation of brown dwarfs and directly imaged exoplanets driven by cloud radiative feedback: effects of rotation. Mon. Not. R. Astron. Soc. 502 , 678–699 (2021).

Parmentier, V., Showman, A. P. & Fortney, J. J. The cloudy shape of hot Jupiter thermal phase curves. Mon. Not. R. Astron. Soc. 501 , 78–108 (2021).

Perna, R., Menou, K. & Rauscher, E. Magnetic drag on hot Jupiter atmospheric winds. Astrophys. J. 719 , 1421–1426 (2010).

Komacek, T. D. & Youdin, A. N. Structure and evolution of internally heated hot Jupiters. Astrophys. J. 844 , 94 (2017).

Rogers, T. M. Constraints on the magnetic field strength of HAT-P-7 b and other hot giant exoplanets. Nat. Astron. 1 , 0131 (2017).

Kataria, T., Showman, A. P., Fortney, J. J., Marley, M. S. & Freedman, R. S. The atmospheric circulation of the super Earth GJ 1214b: dependence on composition and metallicity. Astrophys. J. 785 , 92 (2014).

Zhang, X. & Showman, A. P. Effects of bulk composition on the atmospheric dynamics on close-in exoplanets. Astrophys. J. 836 , 73 (2017).

Carone, L. et al. Equatorial retrograde flow in WASP-43b elicited by deep wind jets? Mon. Not. R. Astron. Soc. 496 , 3582–3614 (2020).

Showman, A. P., Tan, X. & Parmentier, V. Atmospheric dynamics of hot giant planets and brown dwarfs. Space Sci. Rev. 216 , 139 (2020).

Helling, C. et al. Cloud property trends in hot and ultra-hot giant gas planets (WASP-43b, WASP-103b, WASP-121b, HAT-P-7b, and WASP-18b). Astron. Astrophys. 649 , A44 (2021).

Article Google Scholar

Hellier, C. et al. WASP-43b: the closest-orbiting hot Jupiter. Astron. Astrophys. 535 , L7 (2011).

Scandariato, G. et al. Phase curve and geometric albedo of WASP-43b measured with CHEOPS, TESS, and HST WFC3/UVIS. Astron. Astrophys. 668 , A17 (2022).

Stevenson, K. B. et al. Thermal structure of an exoplanet atmosphere from phase-resolved emission spectroscopy. Science 346 , 838–841 (2014).

Stevenson, K. B. et al. Spitzer phase curve constraints for WASP-43b at 3.6 and 4.5 μm. Astron. J. 153 , 68 (2017).

Venot, O. et al. Global chemistry and thermal structure models for the hot Jupiter WASP-43b and predictions for JWST. Astrophys. J. 890 , 176 (2020).

Helling, C. et al. Mineral cloud and hydrocarbon haze particles in the atmosphere of the hot Jupiter JWST target WASP-43b. Astron. Astrophys. 641 , A178 (2020).

Morello, G., Danielski, C., Dickens, D., Tremblin, P. & Lagage, P. O. An independent analysis of the Spitzer/IRAC phase curves of WASP43 b. Astron. J. 157 , 205 (2019).

Murphy, M. M. et al. A lack of variability between repeated Spitzer phase curves of WASP-43b. Astron. J. 165 , 107 (2023).

Wright, G. S. et al. The mid-infrared instrument for jwst and its in-flight performance. Publ. Astron. Soc. Pac. 135 , 048003 (2023).

Kendrew, S. et al. The Mid-Infrared Instrument for the James Webb Space Telescope, IV: the low-resolution spectrometer. Publ. Astron. Soc. Pac. 127 , 623 (2015).

Bouwman, J. et al. Spectroscopic time series performance of the Mid-infrared Instrument on the JWST. Publ. Astron. Soc. Pac. 135 , 038002 (2023).

Esposito, M. et al. The GAPS programme with HARPS-N at TNG. XIII. The orbital obliquity of three close-in massive planets hosted by dwarf K-type stars: WASP-43, HAT-P-20 and Qatar-2. Astron. Astrophys. 601 , A53 (2017).

May, E. M. & Stevenson, K. B. Introducing a new Spitzer master BLISS map to remove the instrument systematic phase-curve-parameter degeneracy, as demonstrated by a reanalysis of the 4.5 μm WASP-43b phase curve. Astron. J. 160 , 140 (2020).

Bell, T. J. et al. A comprehensive reanalysis of Spitzer’s 4.5 μm phase curves, and the phase variations of the ultra-hot Jupiters MASCARA-1b and KELT-16b. Mon. Not. R. Astron. Soc. 504 , 3316–3337 (2021).

Showman, A. P. et al. Atmospheric circulation of hot Jupiters: coupled radiative-dynamical general circulation model simulations of HD 189733b and HD 209458b. Astrophys. J. 699 , 564–584 (2009).

Showman, A. P. & Polvani, L. M. Equatorial superrotation on tidally locked exoplanets. Astrophys. J. 738 , 71 (2011).

Roman, M. & Rauscher, E. Modeling the effects of inhomogeneous aerosols on the hot Jupiter Kepler-7b’s atmospheric circulation. Astrophys. J. 850 , 17 (2017).

Roman, M. & Rauscher, E. Modeled temperature-dependent clouds with radiative feedback in hot Jupiter atmospheres. Astrophys. J. 872 , 1 (2019).

Cowan, N. B. et al. Thermal phase variations of WASP-12b: defying predictions. Astrophys. J. 747 , 82 (2012).

Mendonça, J. M., Tsai, S.-m, Malik, M., Grimm, S. L. & Heng, K. Three-dimensional circulation driving chemical disequilibrium in WASP-43b. Astrophys. J. 869 , 107 (2018).

Agúndez, M., Parmentier, V., Venot, O., Hersant, F. & Selsis, F. Pseudo 2D chemical model of hot-Jupiter atmospheres: application to HD 209458b and HD 189733b. Astron. Astrophys. 564 , A73 (2014).

Tsai, S.-M. et al. Inferring shallow surfaces on sub-Neptune exoplanets with JWST. Astrophys. J. Lett. 922 , L27 (2021).

Moses, J. I., Tremblin, P., Venot, O. & Miguel, Y. Chemical variation with altitude and longitude on exo-Neptunes: predictions for Ariel phase-curve observations. Exp. Astron. 53 , 279–322 (2022).

Baeyens, R., Konings, T., Venot, O., Carone, L. & Decin, L. Grid of pseudo-2D chemistry models for tidally locked exoplanets—II. The role of photochemistry. Mon. Not. R. Astron. Soc. 512 , 4877–4892 (2022).

Gáspár, A. et al. The quantum efficiency and diffractive image artifacts of Si:As IBC mid-IR detector arrays at 5–10 μm: implications for the JWST/MIRI detectors. Publ. Astron. Soc. Pac. 133 , 014504 (2021).

Bell, T. J. et al. A first look at the JWST MIRI/LRS phase curve of WASP-43b. Preprint at https://arxiv.org/abs/2301.06350 (2023).

Bell, T. et al. Eureka!: an end-to-end pipeline for JWST time-series observations. J. Open Source Softw. 7 , 4503 (2022).

Bushouse, H. et al. JWST calibration pipeline. GitHub https://github.com/spacetelescope/jwst (2022).

Schlawin, E. et al. JWST noise floor. I. Random error sources in JWST NIRCam time series. Astron. J. 160 , 231 (2020).

Kempton, E. M.-R. et al. A reflective, metal-rich atmosphere for GJ 1214b from its JWST phase curve. Nature 620 , 67–71 (2023).

Koposov, S. et al. joshspeagle/dynesty: v1.2.3. Zenodo https://doi.org/10.5281/zenodo.6609296 (2022).

Cowan, N. B. & Agol, E. Inverting phase functions to map exoplanets. Astrophys. J. Lett. 678 , L129 (2008).

Cowan, N. B., Chayes, V., Bouffard, É., Meynig, M. & Haggard, H. M. Odd harmonics in exoplanet photometry: weather or artifact? Mon. Not. R. Astron. Soc. 467 , 747–757 (2017).

ADS Google Scholar

Trotta, R. Bayes in the sky: Bayesian inference and model selection in cosmology. Contemp. Phys. 49 , 71–104 (2008).

Welbanks, L. & Madhusudhan, N. Aurora: a generalized retrieval framework for exoplanetary transmission spectra. Astrophys. J. 913 , 114 (2021).

Luger, R. et al. rodluger/starry: v1.2.0. Zenodo https://doi.org/10.5281/zenodo.5567790 (2021).

Bonomo, A. S. et al. The GAPS programme with HARPS-N at TNG. XIV. Investigating giant planet migration history via improved eccentricity and mass determination for 231 transiting planets. Astron. Astrophys. 602 , A107 (2017).

Kipping, D. M. Efficient, uninformative sampling of limb darkening coefficients for two-parameter laws. Mon. Not. R. Astron. Soc. 435 , 2152–2160 (2013).

Salvatier, J., Wiecki, T. V. & Fonnesbeck, C. Probabilistic programming in Python using PyMC3. PeerJ Comput. Sci. 2 , e55 (2016).

Gelman, A. & Rubin, D. B. Inference from iterative simulation using multiple sequences. Stat. Sci. 7 , 457–472 (1992).

Kreidberg, L. batman: basic transit model calculation in Python. Publ. Astron. Soc. Pac. 127 , 1161 (2015).

Foreman-Mackey, D., Hogg, D. W., Lang, D. & Goodman, J. emcee: the MCMC hammer. Publ. Astron. Soc. Pac. 125 , 306 (2013).

Foreman-Mackey, D. et al. exoplanet: gradient-based probabilistic inference for exoplanet data & other astronomical time series. J. Open Source Softw. 6 , 3285 (2021).

Ivshina, E. S. & Winn, J. N. TESS transit timing of hundreds of hot Jupiters. Astrophys. J. Suppl. Ser. 259 , 62 (2022).

Gillon, M. et al. The TRAPPIST survey of southern transiting planets. I. Thirty eclipses of the ultra-short period planet WASP-43 b. Astron. Astrophys. 542 , A4 (2012).

Turbet, M. et al. Day-night cloud asymmetry prevents early oceans on Venus but not on Earth. Nature 598 , 276–280 (2021).

Spiga, A. et al. Global climate modeling of Saturn’s atmosphere. Part II: Multi-annual high-resolution dynamical simulations. Icarus 335 , 113377 (2020).

Charnay, B., Forget, F., Tobie, G., Sotin, C. & Wordsworth, R. Titan’s past and future: 3D modeling of a pure nitrogen atmosphere and geological implications. Icarus 241 , 269–279 (2014).

Turbet, M. et al. The habitability of Proxima Centauri b: II. possible climates and observability. Astron. Astrophys. 596 , A112 (2016).

Charnay, B., Meadows, V., Misra, A., Leconte, J. & Arney, G. 3D modeling of GJ1214b’s atmosphere: formation of inhomogeneous high clouds and observational implications. Astrophys. J. 813 , L1 (2015).

Teinturier, L. et al. The radiative and dynamical impact of clouds in the atmosphere of the hot Jupiter WASP-43 b. Astron. Astrophys. 683 , A231 (2024).

Blain, D., Charnay, B. & Bézard, B. 1D atmospheric study of the temperate sub-Neptune K2-18b. Astron. Astrophys. 646 , A15 (2021).

Falco, A., Zingales, T., Pluriel, W. & Leconte, J. Toward a multidimensional analysis of transmission spectroscopy: I. Computation of transmission spectra using a 1D, 2D, or 3D atmosphere structure. Astron. Astrophys. 658 , A41 (2022).

Adcroft, A., Campin, J.-M., Hill, C. & Marshall, J. Implementation of an atmosphere ocean general circulation model on the expanded spherical cube. Mon. Weather Rev. 132 , 2845 (2004).

Marley, M. S., Gelino, C., Stephens, D., Lunine, J. I. & Freedman, R. Reflected spectra and albedos of extrasolar giant planets. I. Clear and cloudy atmospheres. Astrophys. J. 513 , 879–893 (1999).

Allard, F. & Hauschildt, P. H. Model atmospheres for M (sub)dwarf stars. I. The base model grid. Astrophys. J. 445 , 433 (1995).

Hauschildt, P. H., Allard, F. & Baron, E. The NextGen model atmosphere grid for 3000 ≤ T eff ≤ 10,000 K. Astrophys. J. 512 , 377–385 (1999).

Husser, T. O. et al. A new extensive library of PHOENIX stellar atmospheres and synthetic spectra. Astron. Astrophys. 553 , A6 (2013).

Freedman, R. S., Marley, M. S. & Lodders, K. Line and mean opacities for ultracool dwarfs and extrasolar planets. Astrophys. J. Suppl. Ser. 174 , 504–513 (2008).

Freedman, R. S. et al. Gaseous mean opacities for giant planet and ultracool dwarf atmospheres over a range of metallicities and temperatures. Astrophys. J. Suppl. Ser. 214 , 25 (2014).

Marley, M. S. et al. The Sonora brown dwarf atmosphere and evolution models. I. Model description and application to cloudless atmospheres in rainout chemical equilibrium. Astrophys. J. 920 , 85 (2021).

Visscher, C., Lodders, K. & Fegley, B. Atmospheric chemistry in giant planets, brown dwarfs, and low-mass dwarf stars. III. Iron, magnesium, and silicon. Astrophys. J. 716 , 1060–1075 (2010).

Parmentier, V., Showman, A. P. & Lian, Y. 3D mixing in hot Jupiters atmospheres—I. Application to the day/night cold trap in HD 209458b. Astron. Astrophys. 558 , A91 (2013).

Tan, X. & Showman, A. P. Atmospheric circulation of brown dwarfs and directly imaged exoplanets driven by cloud radiative feedback: global and equatorial dynamics. Mon. Not. R. Astron. Soc. 502 , 2198–2219 (2021).

Komacek, T. D., Tan, X., Gao, P. & Lee, E. K. H. Patchy nightside clouds on ultra-hot Jupiters: general circulation model simulations with radiatively active cloud tracers. Astrophys. J. 934 , 79 (2022).

Ackerman, A. S. & Marley, M. S. Precipitating condensation clouds in substellar atmospheres. Astrophys. J. 556 , 872–884 (2001).

Lee, E. K. et al. 3D radiative transfer for exoplanet atmospheres. gCMCRT: a GPU-accelerated mcrt code. Astrophys. J. 929 , 180 (2022).

Lee, E. K. et al. Simulating gas giant exoplanet atmospheres with exo-fms: comparing semigrey, picket fence, and correlated- k radiative-transfer schemes. Mon. Not. R. Astron. Soc. 506 , 2695–2711 (2021).

Grimm, S. L. et al. Helios-k 2.0 opacity calculator and open-source opacity database for exoplanetary atmospheres. Astrophys. J. Suppl. Ser. 253 , 30 (2021).

Gharib-Nezhad, E. et al. Exoplines: molecular absorption cross-section database for brown dwarf and giant exoplanet atmospheres. Astrophys. J. Suppl. Ser. 254 , 34 (2021).

Schneider, A. D. et al. Exploring the deep atmospheres of HD 209458b and WASP-43b using a non-gray general circulation model. Astron. Astrophys. 664 , A56 (2022).

Mollière, P. et al. petitRADTRANS. a Python radiative transfer package for exoplanet characterization and retrieval. Astron. Astrophys. 627 , A67 (2019).

Chubb, K. L. et al. The ExoMolOP database: cross sections and k -tables for molecules of interest in high-temperature exoplanet atmospheres. Astron. Astrophys. 646 , A21 (2021).

Polyansky, O. L. et al. ExoMol molecular line lists XXX: a complete high-accuracy line list for water. Mon. Not. R. Astron. Soc. 480 , 2597–2608 (2018).

Yurchenko, S. N., Amundsen, D. S., Tennyson, J. & Waldmann, I. P. A hybrid line list for CH 4 and hot methane continuum. Astron. Astrophys. 605 , A95 (2017).

Yurchenko, S. N., Mellor, T. M., Freedman, R. S. & Tennyson, J. ExoMol line lists—XXXIX. Ro-vibrational molecular line list for CO 2 . Mon. Not. R. Astron. Soc. 496 , 5282–5291 (2020).

Coles, P. A., Yurchenko, S. N. & Tennyson, J. ExoMol molecular line lists—XXXV. a rotation–vibration line list for hot ammonia. Mon. Not. R. Astron. Soc. 490 , 4638–4647 (2019).

Li, G. et al. Rovibrational line lists for nine isotopologues of the CO molecule in the X 1 Σ + ground electronic state. Astrophys. J. Suppl. Ser. 216 , 15 (2015).

Azzam, A. A. A., Tennyson, J., Yurchenko, S. N. & Naumenko, O. V. ExoMol molecular line lists—XVI. The rotation-vibration spectrum of hot H 2 S. Mon. Not. R. Astron. Soc. 460 , 4063–4074 (2016).

Barber, R. J. et al. ExoMol line lists—III. \n improved hot rotation-vibration line list for HCN and HNC. Mon. Not. R. Astron. Soc. 437 , 1828–1835 (2014).

Sousa-Silva, C., Al-Refaie, A. F., Tennyson, J. & Yurchenko, S. N. ExoMol line lists—VII. the rotation–vibration spectrum of phosphine up to 1500 K. Mon. Not. R. Astron. Soc. 446 , 2337–2347 (2014).

Wende, S., Reiners, A., Seifahrt, A. & Bernath, P. F. CRIRES spectroscopy and empirical line-by-line identification of FeH molecular absorption in an M dwarf. Astron. Astrophys. 523 , A58 (2010).

Piskunov, N. E., Kupka, F., Ryabchikova, T. A., Weiss, W. W. & Jeffery, C. S. VALD: the Vienna Atomic Line Data Base. Astron. Astrophys. Suppl. 112 , 525 (1995).

Dalgarno, A. & Williams, D. A. Rayleigh scattering by molecular hydrogen. Astrophys. J. 136 , 690–692 (1962).

Chan, Y. M. & Dalgarno, A. The refractive index of helium. Proc. Phys. Soc. 85 , 227–230 (1965).

Richard, C. et al. New section of the HITRAN database: collision-induced absorption (CIA). J. Quant. Spectrosc. Radiat. Transf. 113 , 1276–1285 (2012).

Hoskins, B. & Simmons, A. A multi-layer spectral model and the semi-implicit method. Q. J. R. Meteorol. Soc. 101 , 637–655 (1975).

Menou, K. & Rauscher, E. Atmospheric circulation of hot Jupiters: a shallow three-dimensional model. Astrophys. J. 700 , 887–897 (2009).

Rauscher, E. & Menou, K. Three-dimensional modeling of hot jupiter atmospheric flows. Astrophys. J. 714 , 1334 (2010).

Rauscher, E. & Menou, K. A general circulation model for gaseous exoplanets with double-gray radiative transfer. Astrophys. J. 750 , 96 (2012).

Rauscher, E. & Kempton, E. M. The atmospheric circulation and observable properties of non-synchronously rotating hot Jupiters. Astrophys. J. 790 , 79 (2014).

May, E. M. & Rauscher, E. From super-Earths to mini-Neptunes: implications of a surface on atmospheric circulation. Astrophys. J. 893 , 161 (2020).

Beltz, H. et al. Magnetic drag and 3D effects in theoretical high-resolution emission spectra of ultrahot Jupiters: the case of WASP-76b. Astron. J. 164 , 140 (2022).

Toon, O. B., McKay, C., Ackerman, T. & Santhanam, K. Rapid calculation of radiative heating rates and photodissociation rates in inhomogeneous multiple scattering atmospheres. J. Geophys. Res. Atmos. 94 , 16287–16301 (1989).

Gao, P. et al. Aerosol composition of hot giant exoplanets dominated by silicates and hydrocarbon hazes. Nat. Astron. 4 , 951–956 (2020).

Zhang, J., Kempton, E. M. R. & Rauscher, E. Constraining hot Jupiter atmospheric structure and dynamics through Doppler-shifted emission spectra. Astrophys. J. 851 , 84 (2017).

Malsky, I. et al. Modeling the high-resolution emission spectra of clear and cloudy nontransiting hot Jupiters. Astrophys. J. 923 , 62 (2021).

Mendonça, J. M., Grimm, S. L., Grosheintz, L. & Heng, K. THOR: a new and flexible global circulation model to explore planetary atmospheres. Astrophys. J. 829 , 115 (2016).

Deitrick, R. et al. THOR 2.0: major improvements to the open-source general circulation model. Astrophys. J. Suppl. Ser. 248 , 30 (2020).

Tomita, H. & Satoh, M. A new dynamical framework of nonhydrostatic global model using the icosahedral grid. Fluid Dyn. Res. 34 , 357–400 (2004).

Article ADS MathSciNet Google Scholar

Mendonça, J. M. Mass transport in a moist planetary climate model. Astron. Astrophys. 659 , A43 (2022).

Mendonça, J. M. & Buchhave, L. A. Modelling the 3D climate of Venus with OASIS. Mon. Not. R. Astron. Soc. 496 , 3512–3530 (2020).

Shao, W. D., Zhang, X., Mendonça, J. & Encrenaz, T. Local-time dependence of chemical species in the Venusian mesosphere. Planet. Sci. J. 3 , 3 (2022).

Deitrick, R. et al. The THOR + HELIOS general circulation model: multiwavelength radiative transfer with accurate scattering by clouds/hazes. Mon. Not. R. Astron. Soc. 512 , 3759–3787 (2022).

Tsai, S.-M. et al. Toward consistent modeling of atmospheric chemistry and dynamics in exoplanets: validation and generalization of the chemical relaxation method. Astrophys. J. 862 , 31 (2018).

Malik, M. et al. HELIOS: an open-source, GPU-accelerated radiative transfer code for self-consistent exoplanetary atmospheres. Astron. J. 153 , 56 (2017).

Charnay, B. et al. A survey of exoplanet phase curves with Ariel. Exp. Astron. 53 , 417–446 (2022).

Mendonça, J. M., Read, P. L., Wilson, C. F. & Lee, C. A new, fast and flexible radiative transfer method for Venus general circulation models. Planet. Space Sci. 105 , 80–93 (2015).

Yurchenko, S. N. & Tennyson, J. ExoMol line lists—IV. The rotation–vibration spectrum of methane up to 1500 K. Mon. Not. R. Astron. Soc. 440 , 1649–1661 (2014).

Yurchenko, S. N., Barber, R. J. & Tennyson, J. A variationally computed line list for hot NH 3 . Mon. Not. R. Astron. Soc. 413 , 1828–1834 (2011).

Harris, G. J., Tennyson, J., Kaminsky, B. M., Pavlenko, Y. V. & Jones, H. R. A. Improved HCN/HNC linelist, model atmospheres and synthetic spectra for WZ Cas. Mon. Not. R. Astron. Soc. 367 , 400–406 (2006).

Rothman, L. S. et al. HITEMP, the high-temperature molecular spectroscopic database. J. Quant. Spectrosc. Radiat. Transf. 111 , 2139–2150 (2010).

Rothman, L. S. et al. The HITRAN2012 molecular spectroscopic database. J. Quant. Spectrosc. Radiat. Transf. 130 , 4–50 (2013).

Draine, B. T. Physics of the Interstellar and Intergalactic Medium (Princeton Univ. Press, 2011).

Stock, J. W., Kitzmann, D., Patzer, A. B. C. & Sedlmayr, E. FastChem: a computer program for efficient complex chemical equilibrium calculations in the neutral/ionized gas phase with applications to stellar and planetary atmospheres. Mon. Not. R. Astron. Soc. 479 , 865–874 (2018).

Gandhi, S. & Madhusudhan, N. Retrieval of exoplanet emission spectra with HyDRA. Mon. Not. R. Astron. Soc. 474 , 271–288 (2018).

Feroz, F., Hobson, M. P. & Bridges, M. MULTINEST: an efficient and robust Bayesian inference tool for cosmology and particle physics. Mon. Not. R. Astron. Soc. 398 , 1601–1614 (2009).

Buchner, J. et al. X-ray spectral modelling of the AGN obscuring region in the CDFS: Bayesian model selection and catalogue. Astron. Astrophys. 564 , A125 (2014).

Skilling, J. Nested sampling for general Bayesian computation. Bayesian Anal. 1 , 833–859 (2006).

Article MathSciNet Google Scholar

Gandhi, S., Madhusudhan, N. & Mandell, A. H- and dissociation in ultra-hot Jupiters: a retrieval case study of WASP-18b. Astron. J. 159 , 232 (2020).

Piette, A. A. A. & Madhusudhan, N. Considerations for atmospheric retrieval of high-precision brown dwarf spectra. Mon. Not. R. Astron. Soc. 497 , 5136–5154 (2020).

Piette, A. A. A., Madhusudhan, N. & Mandell, A. M. HyDRo: atmospheric retrieval of rocky exoplanets in thermal emission. Mon. Not. R. Astron. Soc. 511 , 2565–2584 (2022).

Gandhi, S., Madhusudhan, N., Hawker, G. & Piette, A. HyDRA-H: simultaneous hybrid retrieval of exoplanetary emission spectra. Astron. J. 158 , 228 (2019).

Gandhi, S. et al. Spatially resolving the terminator: variation of Fe, temperature, and winds in WASP-76 b across planetary limbs and orbital phase. Mon. Not. R. Astron. Soc. 515 , 749–766 (2022).

Madhusudhan, N. & Seager, S. A temperature and abundance retrieval method for exoplanet atmospheres. Astrophys. J. 707 , 24–39 (2009).

Taylor, J. et al. Understanding and mitigating biases when studying inhomogeneous emission spectra with JWST. Mon. Not. R. Astron. Soc. 493 , 4342–4354 (2020).

Yurchenko, S. N., Tennyson, J., Barber, R. J. & Thiel, W. Vibrational transition moments of CH 4 from first principles. Journal of Molecular Spectroscopy 291 , 69–76 (2013).

Gordon, I. E. et al. The HITRAN2016 molecular spectroscopic database. J. Quant. Spectrosc. Radiat. Transf. 203 , 3–69 (2017).

Underwood, D. S. et al. ExoMol molecular line lists—XIV. The rotation–vibration spectrum of hot SO 2 . Mon. Not. R. Astron. Soc. 459 , 3890–3899 (2016).

Chubb, K. L. et al. Marvel analysis of the measured high-resolution rovibrational spectra of H 2 32 S. J. Quant. Spectrosc. Radiat. Transf. 218 , 178–186 (2018).

Pinhas, A. & Madhusudhan, N. On signatures of clouds in exoplanetary transit spectra. Mon. Not. R. Astron. Soc. 471 , 4355–4373 (2017).

Cubillos, P. E. & Blecic, J. The PyratBay framework for exoplanet atmospheric modeling: a population study of Hubble/WFC3 transmission spectra. Mon. Not. R. Astron. Soc. 505 , 2675–2702 (2021).

Burrows, A., Marley, M. S. & Sharp, C. M. The near-infrared and optical spectra of methane dwarfs and brown dwarfs. Astrophys. J. 531 , 438–446 (2000).

Kurucz, R. L. ATLAS: A Computer Program for Calculating Model Stellar Atmospheres SAO Special Report 309 (SAO, 1970).

Lecavelier Des Etangs, A., Pont, F., Vidal-Madjar, A. & Sing, D. Rayleigh scattering in the transit spectrum of HD 189733b. Astron. Astrophys. 481 , L83–L86 (2008).

Tennyson, J. et al. The ExoMol database: molecular line lists for exoplanet and other hot atmospheres. J. Mol. Spectrosc. 327 , 73–94 (2016).

Cubillos, P. E. An algorithm to compress line-transition data for radiative-transfer calculations. Astrophys. J. 850 , 32 (2017).

Borysow, A., Jorgensen, U. G. & Fu, Y. High-temperature (1000–7000 K) collision-induced absorption of H 2 pairs computed from the first principles, with application to cool and dense stellar atmospheres. J. Quant. Spectrosc. Radiat. Transf. 68 , 235–255 (2001).

Parmentier, V. & Guillot, T. A non-grey analytical model for irradiated atmospheres. I. Derivation. Astron. Astrophys. 562 , A133 (2014).

Blecic, J., Harrington, J. & Bowman, M. O. TEA: a code calculating thermochemical equilibrium abundances. Astrophys. J. Suppl. Ser. 225 , 4 (2016).

Line, M. R. & Parmentier, V. The influence of nonuniform cloud cover on transit transmission spectra. Astrophys. J. 820 , 78 (2016).

Benneke, B. Strict upper limits on the carbon-to-oxygen ratios of eight hot Jupiters from self-consistent atmospheric retrieval. Preprint at https://arxiv.org/abs/1504.07655 (2015).

Benneke, B. & Seager, S. How to distinguish between cloudy mini-Neptunes and water/volatile-dominated super-Earths. Astrophys. J. 778 , 153 (2013).

Morley, C. V. et al. Neglected clouds in T and Y dwarf atmospheres. Astrophys. J. 756 , 172 (2012).

ter Braak, C. J. F. & Vrugt, J. A. Differential evolution Markov chain with snooker updater and fewer chains. Stat. Comput. 18 , 435–446 (2008).

Cubillos, P. et al. On correlated-noise analyses applied to exoplanet light curves. Astron. J. 153 , 3 (2017).

Irwin, P. G. J. et al. The NEMESIS planetary atmosphere radiative transfer and retrieval tool. J. Quant. Spectrosc. Radiat. Transf. 109 , 1136–1150 (2008).

Krissansen-Totton, J., Garland, R., Irwin, P. & Catling, D. C. Detectability of biosignatures in anoxic atmospheres with the James Webb Space Telescope: a TRAPPIST-1e case study. Astron. J. 156 , 114 (2018).

Lacis, A. A. & Oinas, V. A description of the correlated- k distribution method for modelling nongray gaseous absorption, thermal emission, and multiple scattering in vertically inhomogeneous atmospheres. J. Geophys. Res. 96 , 9027–9064 (1991).

Barstow, J. K. Unveiling cloudy exoplanets: the influence of cloud model choices on retrieval solutions. Mon. Not. R. Astron. Soc. 497 , 4183–4195 (2020).

Irwin, P. G. J. et al. 2.5D retrieval of atmospheric properties from exoplanet phase curves: application to WASP-43b observations. Mon. Not. R. Astron. Soc. 493 , 106–125 (2020).

Borysow, A. Collision-induced absorption coefficients of H 2 pairs at temperatures from 60 K to 1000 K. Astron. Astrophys. 390 , 779–782 (2002).

Benneke, B. et al. Water vapor and clouds on the habitable-zone sub-Neptune exoplanet K2-18b. Astrophys. J. Lett. 887 , L14 (2019).

Pelletier, S. et al. Where is the water? Jupiter-like C/H ratio but strong H 2 O depletion found on τ Boötis b using SPIRou. Astron. J. 162 , 73 (2021).

Stock, J. W., Kitzmann, D. & Patzer, A. B. C. FASTCHEM 2: an improved computer program to determine the gas-phase chemical equilibrium composition for arbitrary element distributions. Mon. Not. R. Astron. Soc. 517 , 4070–4080 (2022).

Bell, K. L. & Berrington, K. A. Free-free absorption coefficient of the negative hydrogen ion. J. Phys. B 20 , 801–806 (1987).

John, T. L. Continuous absorption by the negative hydrogen ion reconsidered. Astron. Astrophys. 193 , 189–192 (1988).

Chubb, K. L., Tennyson, J. & Yurchenko, S. N. ExoMol molecular line lists—XXXVII. spectra of acetylene. Mon. Not. R. Astron. Soc. 493 , 1531–1545 (2020).

McKemmish, L. K. et al. ExoMol molecular line lists—XXXIII. The spectrum of titanium oxide. Mon. Not. R. Astron. Soc. 488 , 2836–2854 (2019).

McKemmish, L. K., Yurchenko, S. N. & Tennyson, J. ExoMol line lists—XVIII. The high-temperature spectrum of VO. Mon. Not. R. Astron. Soc. 463 , 771–793 (2016).

Zhang, M., Chachan, Y., Kempton, E. M. R., Knutson, H. A. & Chang, W. H. PLATON II: new capabilities and a comprehensive retrieval on HD 189733b transit and eclipse data. Astrophys. J. 899 , 27 (2020).

Line, M. R. & Yung, Y. L. A systematic retrieval analysis of secondary eclipse spectra. III. Diagnosing chemical disequilibrium in planetary atmospheres. Astrophys. J. 779 , 3 (2013).

Ormel, C. W. & Min, M. ARCiS framework for exoplanet atmospheres—the cloud transport model. Astron. Astrophys. 622 , A121 (2019).

Min, M., Ormel, C. W., Chubb, K., Helling, C. & Kawashima, Y. The ARCiS framework for exoplanet atmospheres: modelling philosophy and retrieval. Astron. Astrophys. 642 , A28 (2020).

Chubb, K. L., Min, M., Kawashima, Y., Helling, C. & Waldmann, I. Aluminium oxide in the atmosphere of hot Jupiter WASP-43b. Astron. Astrophys. 639 , A3 (2020).

Kreidberg, L. et al. A Precise water abundance measurement for the hot Jupiter WASP-43b. Astrophys. J. Lett. 793 , L27 (2014).

Chubb, K. L. & Min, M. Exoplanet atmosphere retrievals in 3D using phase curve data with ARCiS: application to WASP-43b. Astron. Astrophys. 665 , A2 (2022).

Blecic, J. et al. Spitzer observations of the thermal emission from WASP-43b. Astrophys. J. 781 , 116 (2014).

Patrascu, A. T., Tennyson, J. & Yurchenko, S. N. ExoMol molecular linelists: VII: the spectrum of AlO. Mon. Not. R. Astron. Soc. 449 , 3613–3619 (2015).

Tennyson, J. et al. The 2020 release of the ExoMol database: molecular line lists for exoplanet and other hot atmospheres. J. Quant. Spectrosc. Radiat. Transf. 255 , 107228 (2020).

Barton, E. J., Yurchenko, S. N. & Tennyson, J. ExoMol molecular linelists—II. The ro-vibrational spectrum of SiO. Mon. Not. R. Astron. Soc. 434 , 1469–1475 (2013).

Line, M. R., Teske, J., Burningham, B., Fortney, J. J. & Marley, M. S. Uniform atmospheric retrieval analysis of ultracool dwarfs. I. Characterizing benchmarks, GI570D and HD3651b. Astrophys. J. 807 , 183 (2015).

Guillot, T. On the radiative equilibrium of irradiated planetary atmospheres. Astron. Astrophys. 520 , A27 (2010).

Bell, T. J. et al. Products and models for "Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b". Zenodo https://doi.org/10.5281/zenodo.10525170 (2024).

Malik, M. et al. Self-luminous and irradiated exoplanetary atmospheres explored with HELIOS. Astron. J. 157 , 170 (2019).

Harris, C. R. et al. Array programming with NumPy. Nature 585 , 357–362 (2020).

Astropy Collaboration. Astropy: a community Python package for astronomy. Astron. Astrophys. 558 , A33 (2013).

Astropy Collaboration. The Astropy Project: building an open-science project and status of the v2.0 core package. Astron. J. 156 , 123 (2018).

Hunter, J. D. Matplotlib: a 2D graphics environment. Comput. Sci. Eng. 9 , 90–95 (2007).

Download references

Acknowledgements

T.J.B. acknowledges funding support from the NASA Next Generation Space Telescope Flight Investigations program (now JWST) via WBS 411672.07.05.05.03.02. J.K.B. is supported by a UKRI/STFC Ernest Rutherford Fellowship (grant ST/T004479/1). J.B. acknowledges the support received in part from the NYUAD IT High Performance Computing resources, services, and staff expertise. E.D. acknowledges funding as a Paris Region Fellow through the Marie Sklodowska-Curie Action. M.Z. and B.V.R. acknowledge funding from the 51 Pegasi b Fellowship. A.D.F. acknowledges support from the NSF Graduate Research Fellowship Program. M.M., D.P. and L.W. acknowledge funding from the NHFP Sagan Fellowship Program. P.E.C. is funded by the Austrian Science Fund (FWF) Erwin Schroedinger Fellowship program J4595-N. K.L.C. acknowledges funding from STFC, under project number ST/V000861/1. L.D. acknowledges funding from the KU Leuven Interdisciplinary Grant (IDN/19/028), the European Union H2020-MSCA-ITN-2019 under grant no. 860470 (CHAMELEON) and the FWO research grant G086217N. O.V. acknowledges funding from the ANR project ‘EXACT’ (ANR-21-CE49-0008-01) and from the Centre National d’Études Spatiales (CNES). L.T. and B.C. acknowledge access to the HPC resources of MesoPSL financed by the Region Ile de France and the project Equip@Meso (reference ANR-10-EQPX-29-01) of the programme Investissements d’Avenir supervised by the Agence Nationale pour la Recherche.

Author information

Authors and affiliations.

BAER Institute, NASA Ames Research Center, Moffet Field, CA, USA

Taylor J. Bell

Space Science and Astrobiology Division, NASA Ames Research Center, Moffett Field, CA, USA

Leiden Observatory, University of Leiden, Leiden, The Netherlands

Nicolas Crouzet & Sebastian Zieba

INAF - Osservatorio Astrofisico di Torino, Turin, Italy

Patricio E. Cubillos

Space Research Institute, Austrian Academy of Sciences, Graz, Austria

Patricio E. Cubillos, Ludmila Carone & Christiane Helling

Max Planck Institute for Astronomy, Heidelberg, Germany

Laura Kreidberg, Luigi Mancini, Thomas M. Evans-Soma, Maria E. Steinrueck & Sebastian Zieba

Earth and Planets Laboratory, Carnegie Institution for Science, Washington DC, USA

Anjali A. A. Piette, Peter Gao & Munazza K. Alam

School of Physics and Astronomy, University of Leicester, Leicester, UK

Michael T. Roman & Sarah L. Casewell

Universidad Adolfo Ibáñez: Peñalolén, Santiago, Chile

Michael T. Roman

School of Physical Sciences, The Open University, Milton Keynes, UK

Joanna K. Barstow

Department of Physics, New York University Abu Dhabi, Abu Dhabi, United Arab Emirates

Jasmina Blecic & Ian Dobbs-Dixon

Center for Astro, Particle and Planetary Physics (CAP3), New York University Abu Dhabi, Abu Dhabi, United Arab Emirates

Jasmina Blecic

Department of Physics and Trottier Institute for Research on Exoplanets, Université de Montréal, Montreal, Quebec, Canada

Louis-Philippe Coulombe, Björn Benneke, Caroline Piaulet & Jake Taylor

Université Paris-Saclay, Université Paris Cité, CEA, CNRS, AIM, Gif-sur-Yvette, France

Elsa Ducrot, Achrène Dyrek & Pierre-Olivier Lagage

Atmospheric, Oceanic and Planetary Physics, Department of Physics, University of Oxford, Oxford, UK

Mark Hammond, Xianyu Tan & Jake Taylor

DTU Space, Technical University of Denmark, Kongens Lyngby, Denmark

João M. Mendonça

Space Science Institute, Boulder, CO, USA

Julianne I. Moses

Université Côte d’Azur, Observatoire de la Côte d’Azur, CNRS Laboratoire Lagrange, Nice, France

Vivien Parmentier

Johns Hopkins APL, Laurel, MD, USA

Kevin B. Stevenson

LESIA, Observatoire de Paris, Université PSL, Sorbonne Université, Université Paris Cité, CNRS, Meudon, France

Lucas Teinturier & Benjamin Charnay

Laboratoire de Météorologie Dynamique, IPSL, CNRS, Sorbonne Université, Ecole Normale Supérieure, Université PSL, Ecole Polytechnique, Institut Polytechnique de Paris, Paris, France

Lucas Teinturier

Department of Astronomy and Astrophysics, University of Chicago, Chicago, IL, USA

Michael Zhang, Jacob L. Bean & Adina D. Feinstein

Department of Astronomy and Astrophysics, University of California, Santa Cruz, Santa Cruz, CA, USA

Natalie M. Batalha & Aarynn L. Carter

Centre for Exoplanet Science, University of St Andrews, St Andrews, UK

Katy L. Chubb

Center for Space and Habitability, University of Bern, Bern, Switzerland

Brice-Olivier Demory & Elspeth K. H. Lee

Space and Planetary Sciences, Institute of Physics, University of Bern, Bern, Switzerland

Brice-Olivier Demory

Center for Astrophysics | Harvard & Smithsonian, Cambridge, MA, USA

Mercedes López-Morales & Diana Powell

Instituto de Astrofsica de Canarias (IAC), Tenerife, Spain

Giuseppe Morello & Enric Palle

INAF- Palermo Astronomical Observatory, Piazza del Parlamento, Palermo, Italy

Giuseppe Morello

Department of Space, Earth and Environment, Chalmers University of Technology, Gothenburg, Sweden

Department of Astronomy, University of Michigan, Ann Arbor, MI, USA

Emily Rauscher, Ryan C. Challener & Isaac Malsky

Department of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, MD, USA

David K. Sing

Department of Physics and Astronomy, Johns Hopkins University, Baltimore, MD, USA

David K. Sing & Néstor Espinoza

Tsung-Dao Lee Institute, Shanghai Jiao Tong University, Shanghai, People’s Republic of China

School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai, People’s Republic of China

Université Paris Cité and Univ Paris Est Creteil, CNRS, LISA, Paris, France

Olivia Venot

School of Physics, University of Bristol, Bristol, UK

Hannah R. Wakeford

Indian Institute of Technology, Indore, India

Keshav Aggarwal

Centre for Exoplanets and Habitability, University of Warwick, Coventry, UK

Eva-Maria Ahrer

Department of Physics, University of Warwick, Coventry, UK

Anton Pannekoek Institute for Astronomy, University of Amsterdam, Amsterdam, The Netherlands

Robin Baeyens & Jean-Michel Désert

Departamento de Astrofsica, Centro de Astrobiologa (CAB, CSIC-INTA), ESAC campus, Madrid, Spain

David Barrado

Instituto de Astrofisica, Facultad Ciencias Exactas, Universidad Andres Bello, Santiago, Chile

Claudio Caceres

Centro de Astrofisica y Tecnologias Afines (CATA), Santiago, Chile

Nucleo Milenio de Formacion Planetaria (NPF), Valparaíso, Chile

Department of Physics and Astronomy, University of Kansas, Lawrence, KS, USA

Ian J. M. Crossfield

Institute of Astronomy, Department of Physics and Astronomy, KU Leuven, Leuven, Belgium

Leen Decin & Aaron D. Schneider

Space Telescope Science Institute, Baltimore, MD, USA

Néstor Espinoza & Nikolay K. Nikolov

Department of Astrophysical and Planetary Sciences, University of Colorado Boulder, Boulder, CO, USA

Adina D. Feinstein

School of Physics, Trinity College Dublin, Dublin, Ireland

Neale P. Gibson

Planetary Sciences Group, Department of Physics and Florida Space Institute, University of Central Florida, Orlando, FL, USA

Joseph Harrington

Astrophysics Section, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA

Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA, USA

Institute of Planetary Research—Extrasolar Planets And Atmospheres, German Aerospace Center (DLR), Berlin, Germany

Nicolas Iro

Department of Astronomy, University of Maryland, College Park, MD, USA

Eliza M.-R. Kempton, Thaddeus D. Komacek & Matthew C. Nixon

European Space Agency, Space Telescope Science Institute, Baltimore, MD, USA

Sarah Kendrew

California Institute of Technology, IPAC, Pasadena, CA, USA

Jessica Krick

Laboratoire d’Astrophysique de Bordeaux, Université de Bordeaux, Pessac, France

Jérémy Leconte

Département d’Astronomie, Université de Genève, Sauverny, Switzerland

Monika Lendl & Dominique J. M. Petit dit de la Roche

Department of Mathematics and Statistics, University of Exeter, Exeter, UK

Neil T. Lewis

Department of Physics, Utah Valley University, Orem, UT, USA

Joshua D. Lothringer

Department of Physics, University of Rome “Tor Vergata”, Rome, Italy

Luigi Mancini

INAF - Turin Astrophysical Observatory, Turin, Italy

Steward Observatory, University of Arizona, Tucson, AZ, USA

Megan Mansfield

Department of Physics and Astronomy, Faculty of Environment, Science and Economy, University of Exeter, Exeter, UK

Nathan J. Mayne

School of Information and Physical Sciences, University of Newcastle, Callaghan, NSW, Australia

Thomas M. Evans-Soma

Universitäts-Sternwarte, Ludwig-Maximilians-Universität München, Munich, Germany

Karan Molaverdikhani

Exzellenzcluster Origins, Garching, Germany

Department of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA, USA

Benjamin V. Rackham

Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA, USA

Centre for ExoLife Sciences, Niels Bohr Institute, Copenhagen, Denmark

Aaron D. Schneider

School of Earth and Space Exploration, Arizona State University, Tempe, AZ, USA

Luis Welbanks

Department of Physics and Astronomy, University College London, London, UK

Sergei N. Yurchenko

Department of Earth and Planetary Sciences, University of California, Santa Cruz, Santa Cruz, CA, USA

You can also search for this author in PubMed Google Scholar

Contributions

All authors played an appreciable role in one or more of the following: development of the original proposal, management of the project, definition of the target list and observation plan, analysis of the data, theoretical modelling, and preparation of this paper. Some specific contributions are listed as follows. N.M.B., J.L.B. and K.B.S. provided overall programme leadership and management. L.K. and N.C. coordinated the MIRI working group. L.K., V.P., K.B.S., D.K.S., E.M.-R.K., O.V. and P.E.C. made substantial contributions to the design of the programme and the observing proposal. K.B.S. generated the observing plan with input from the team. A.D., P.-O.L., R.C.C., A.L.C., G.M. and M.M. led or co-led working groups and/or contributed to important strategic planning efforts like the design and implementation of the pre-launch data challenges. P.E.C., D.K.S., R.C.C., P.-O.L. and J.B. generated simulated data for pre-launch testing of methods. L.K., T.J.B., M.T.R., N.C., V.P., A.A.A.P. and J.I.M. contributed substantially to the writing of this paper. T.J.B., N.C., M.Z. and E.D. contributed to the development of data analysis pipelines and/or provided the data analysis products used in this analysis that is, reduced the data, modelled the light curves, and/or produced the planetary spectrum. A.A.A.P. coordinated the atmospheric retrieval analysis with contributions from J.K.B., J.B., L.-P.C., M.Z., and K.L.C. M.T.R. coordinated the GCM results and interpretation with contributions from X.T., L.T., L.C., J.M.M. and I.M. T.J.B., N.C., P.E.C., J.B., L.-P.C. and M.H. generated figures for this paper. M.C.N., X.Z., B.V.R., J.K., M.L.-M., B.C., S.L.C. and R.H. provided substantial feedback to the paper, and G.M. and K.L.C. coordinated comments from all authors.

Corresponding author

Correspondence to Taylor J. Bell .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Peer review

Peer review information.

Nature Astronomy thanks the anonymous reviewer(s) for their contribution to the peer review of this work

Additional information

Publisher ’ s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended data fig. 1 the underestimation of uncertainties as a function of spectral binning for the l168-9b commissioning observations..

a , The observed L168-9b transmission spectrum with 1 σ error bars for spectrally unbinned data (grey circles), 0.15 μ m bins (black squares), 0.5 μ m bins (large red circles), and a 5-12 μ m broadband bin (horizontal blue shaded region). The spectrum for wavelength pairs is not shown to avoid excessive clutter. b , The median of the transit depth uncertainties are shown with blue squares, while the observed scatter in the transmission spectrum is shown with orange circles. For unbinned data, the transmission spectrum shows about 2.5 × the scatter predicted by the fits to the individual light curves. Binning pairs of wavelengths reduces the level of underestimation of the scatter in the transmission spectrum, but considerable excess noise remains. Coarser binning schemes like the constant 0.15 μ m bins used in the MIRI time-series observation commissioning paper 29 or the 0.5 μ m bins we use in this work further reduce the level of uncertainty underestimation.

Extended Data Fig. 2 A model-independent demonstration of the initial changes in flux for the WASP-43b observations.

a , The first 120 minutes of three of our spectroscopically binned light curves of WASP-43b (with 1 σ uncertainties) showing the initial settling behaviour as a function of wavelength. A teal dashed line shows the amplitude of a -0.25% change in flux compared to the values around 120 minutes, and a magenta dotted line shows a +0.25% change. b , A summary of the ramp amplitudes, signs, and timescales for each of our wavelength bins (with 1 σ uncertainties). The teal and magenta horizontal lines are the same as those in panel a to aid in translating between the two figures. At short wavelengths, the flux sharply drops by about 0.5% within the first 30 minutes and then largely settles but does continue to decrease with time. With increasing wavelength, the strength of this initial ramp decreases and eventually changes sign, becoming an upwards ramp. Within the ‘shadowed region’ (marked in red), the light curves show a very strong upwards ramp that takes much longer (greater than about 60 minutes) to appreciably decay. It is important to note that the data in this figure also includes a small amount of astrophysical phase variations which should result in a small increase in flux of less than 0.05% per hour.

Extended Data Fig. 3 Retrieved spectra from the six retrievals.

a , Median retrieved nightside spectra for the HyDRA (dark blue line), NEMESIS (dash-dotted gold line), and PyratBay (dashed magenta line) and their 1 σ contour. The regions of higher water opacity are indicated by the purple shading at the top of the panel, with the observed rise in flux at 6.3 μ m being caused by a drop in opacity. b , c , and d , Same as panel a for the evening terminator, dayside, and morning terminator respectively. e , f , g , and h , Same as panels a, b, c, and d, for the SCARLET (dashed red line), PLATON (blue line), and ARCiS (dash-dotted green line) retrievals.

Extended Data Fig. 4 Chemically-consistent atmospheric retrievals.

Same as Figure 4 but for retrievals assuming thermochemical-equilibrium abundances consistent with the pressure-temperature profiles. a , 1 σ credible interval contours of the temperature profiles. The black curves show the predicted temperature profile from a 2D radiative-transport model46. The vertical bars show the range of pressures probed by the observations. b and c , probability posterior distributions for H 2 O and CH 4 abundances, respectively. The shaded area for each curve denotes the 1 σ credible interval of each posterior. The green and blue bars denote the abundances predicted by equilibrium and disequilibrium-chemistry models with solar abundances, respectively, at the pressures probed by the observations. Compared to the free-chemistry retrievals, the thermochemical-equilibrium retrievals on the nightside spectra produced worse fits, this is driven particularly by the higher amount of methane expected under equilibrium chemistry.

Extended Data Fig. 5 Retrieval contribution functions.

Contribution functions integrated over the data point spectral bins, at each phase ( a-d ), and for each retrieval framework. These curves show the range of pressures probed by the observation according to the atmospheric models. The enhanced opacity from the water band around 7-9 μ m makes these wavelengths probe lower pressures and hence colder temperatures, whereas the rest of the observing window probes higher pressures and higher temperatures.

Extended Data Fig. 6 PyratBay clouds exploration.

a , Cloud species that condense in the temperature regime expected for the WASP-43b nightside. Dashed lines represent vapour pressure curves 162 for each species assuming solar composition, while the coloured ranges denote the corresponding extent of the vapour pressure curves assuming 100 × sub- and super-solar atmospheric composition. The extent of the retrieved nightside contribution functions is shown in grey, and the extent of the retrieved temperature uncertainties is shown in light purple. The intersection between the contribution function and temperature ranges indicates the pressures at which we could observe cloud condensation and potentially detect their spectral features, if present in the observations. b , Panels display the retrieved posterior density plots for the explored cloud parameters of the TSC model (cloud number density, q*; effective particle size, r eff ; and the standard deviation of the log-normal distribution, ${\sigma }_{\log }$ ) for the MnS clouds. The black vertical line denotes the parameter’s median value, while the extent of the purple region denotes the 1 σ uncertainties, both given at the top left corner of the panel. Similar, fully non-constrained posteriors are retrieved for other explored cloud species, MgSiO 3 , ZnS, and KCl, suggesting the lack of observable spectral characteristics from clouds in the observed data.

Extended Data Fig. 7 A comparison of the retrieved temperature-pressure profiles to the GCM simulations.

Each of a-d shows the temperature profile retrieved by HyDRA, compared to the GCM simulations highlighted in Figure 3 and listed in Extended Data Table 1. The GCM temperature profiles are calculated at phases 0.0, 0.25, 0.5, and 0.75 by averaging over the visible hemisphere by viewing angle, to produce a one-dimensional profile that is comparable to the retrieved profile. The GCM simulations are generally warmer on the nightside than the retrieved temperatures; cloudy simulations emit from lower pressures and so match the observed lower brightness temperatures better (see the contribution functions in Extended Data Fig. 5).

Supplementary information

Supplementary information.

Supplementary Table 1 and Figs. 1 and 2.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Bell, T.J., Crouzet, N., Cubillos, P.E. et al. Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b. Nat Astron (2024). https://doi.org/10.1038/s41550-024-02230-x

Download citation

Received : 01 November 2023

Accepted : 22 February 2024

Published : 30 April 2024

DOI : https://doi.org/10.1038/s41550-024-02230-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

Explore articles by subject
Guide to authors
Editorial policies

Smartphone 360
Smartphone Model Tracker
Refurbished Market Tracker
Smartwatch 360
Emerging Tech
Consumer Research
Smartphone AP-SoC
Teardown & BOM
Cellular Tracker & Forecasts
Smart home Tracker
PCs and Tablets
5G Networks
Press releases
White Papers
Events & Webinars
Counterpoint Conversations
Research Portal
Research services
Marketing engagement
Semis professionals
Mobile operators
Smartphone professionals
Report purchases
Software and services
Opinion leaders
In the press
Vendor briefings

Global Smartphone Market Grows 6% YoY in Q1 2024; Revenue Reaches Highest Level in a First Quarter

/ May 3, 2024
/ Team Counterpoint
The global smartphone market grew by 6% YoY to reach 296.9 million unit shipments in Q1 2024.
Samsung dethroned Apple to become the top smartphone player globally, accounting for 20% shipment share.
Apple’s shipments declined 13% YoY due to high inventory build-up during the last quarter. However, the brand’s ASP grew 2% YoY.
Among the top five OEMs, Xiaomi grew the fastest, registering 34% YoY shipment growth.
Global smartphone revenues also grew 7% YoY and reached highest-ever for a first quarter.

London, New Delhi, Hong Kong, Seoul, Beijing, Denver, Buenos Aires – May 3, 2024

Global smartphone shipments increased 6% YoY in the first quarter of 2024, according to the latest research from Counterpoint’s Market Monitor service. The growth is mostly attributed to the strong performance in key regions such as Europe, Middle East and Africa (MEA) and Caribbean and Latin America (CALA).

Commenting on overall market dynamics, Senior Analyst Prachir Singh said, “Smartphone shipments grew during the first quarter as emerging markets continued their strong momentum, while Europe , especially Central and Eastern Europe, grew the most compared to a difficult Q1 2023. Consumer demand in these markets has been growing gradually and the inventory levels have improved. MEA was the fastest growing region due to the strong shipments of TECNO, Xiaomi and HONOR . China also grew YoY owing to strong Lunar New Year sales and Huawei’s comeback. The India market also saw growth thanks to exiting 2023 with healthy inventory levels. The mature markets of North America and Japan declined relative to the same period in 2023.”

Global smartphone revenues grew by 7% YoY in Q1 2024 and reached the highest-ever in a first calendar quarter. The >$800 price segment was the fastest growing, registering double-digit growth and accounting for 18% of smartphone shipments in Q1 2024, up by 2pp compared to Q1 2023. Apple led the smartphone market revenues with a 43% share, although its revenues declined by 11% YoY. Samsung’s revenues grew 2% YoY, propelled by its increasing ASP while shipments remained flat. Among the top five OEMs, Xiaomi’s revenue growth was the fastest due to strong performance in its key markets. Revenues for the market beyond the top five OEMs also grew significantly, driven by increasing revenues from Huawei, HONOR and Transsion brands.

Commenting on Apple’s performance, Research Director Jeff Fieldhack said, “Tough competition in China, record low upgrades in the US and a difficult compare from last year due to iPhone 14 Pro’s supply shifting to Q1 2023 all weighed on iPhone performance. However, there were upsides as well. An improved product mix with 15 Pro’s performing better than its predecessors, and an increasing footprint in emerging markets, helped Apple in arresting some of the declines. Emerging markets especially provide long term growth opportunities. We also expect the inclusion of GenAI later this year to contribute to iPhone upgrades.”

Samsung re-captured the top spot and led global smartphone shipments during the quarter, driven by the early refresh of the Galaxy-A-series and strong performance of the Galaxy S24 series. Samsung reached its highest-ever ASP during the quarter as well. Among the biggest smartphone brands from China, Xiaomi and vivo registered growth. While Xiaomi registered growth in almost all its key markets, vivo posted strong performance in emerging markets of Asia Pacific including India. Huawei , HONOR and Transsion were the other key brands that gained during the quarter. Huawei made huge gains in China and won share from the leading brands. HONOR gained in CALA and MEA, in addition to having a strong share in China. The Transsion brands, TECNO, itel and Infinix, performed strongly in APAC, Eastern Europe, India and the MEA. OPPO* experienced a shipment decline due to stiff competition in key markets like China. The brand also prioritized value over shipments, with a shift to pushing higher ASP devices in emerging markets.

Commenting on the near-term outlook, Research Director Tarun Pathak said, “Growth is expected to be slow but steady in the near term. However, revenues are expected to grow faster as the ongoing premiumization trend is likely to persist, especially with the rise of newer form factors and capabilities such as foldables and GenAI . More than 10 OEMs have launched over 30 GenAI-capable smartphones so far. We estimate that GenAI’s share of overall smartphone shipments will reach 11% by 2024.”

*OPPO includes OnePlus since Q3 2021

Note: Pricing analysis is based on wholesale prices.

Feel free to reach us at [email protected] for questions regarding our latest research and insights.

Counterpoint Technology Market Research is a global research firm specializing in products in the TMT (technology, media and telecom) industry. It services major technology and financial firms with a mix of monthly reports, customized projects and detailed analyses of the mobile and technology markets. Its key analysts are seasoned experts in the high-tech

Follow Counterpoint Research

press(at)counterpointresearch.com

Apple Could be Looking for iPhone GenAI Partnerships in China and Globally
Apple's Services Revenue Grows to Reach Highest Ever Share as Product Revenues Decline
China Smartphone Sales Remain in Black on Huawei, HONOR, Xiaomi Outperformance
GenAI-capable Smartphone Shipments to Grow Over 4X by 2027
Global Smartphone Shipments to See Modest Rebound in 2024 Driven by Premium, Budget-Economy Segments
Qualcomm Snapdragon 8s Gen 3 to Promote AI Smartphones

Only fill in if you are not human

Teardown & BOM
Cellular Tracker & Forecasts
Events & Webinars

Term of Use and Privacy Policy

Help | Advanced Search

Computer Science > Computation and Language

Title: a primer on the inner workings of transformer-based language models.

Abstract: The rapid progress of research aimed at interpreting the inner workings of advanced language models has highlighted a need for contextualizing the insights gained from years of work in this area. This primer provides a concise technical introduction to the current techniques used to interpret the inner workings of Transformer-based language models, focusing on the generative decoder-only architecture. We conclude by presenting a comprehensive overview of the known internal mechanisms implemented by these models, uncovering connections across popular approaches and active research directions in this area.

Submission history

Access paper:.

Other Formats

References & Citations

Google Scholar
Semantic Scholar

BibTeX formatted citation

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

Centre for Social Research & Methods

Education & training
Programs & research
Working papers
Methods research papers
COVID-19 publications
Other publications

Related Sites

ANU College of Arts & Social Sciences
Research School of Social Sciences
Australian National Internships Program

It’s the economy (and housing), stupid: Views of Australians on the economy and the housing market in January 2024

This paper summarises recent public opinion data on views on the economy, as well as experience and views on the housing market. It is based on data from the January 2024 ANUpoll as well as historic ANUpoll data to show that compared to almost any time during the pandemic or immediately beforehand, Australians are more financially stressed. They are also less satisfied with the direction of the country, and less satisfied with their own life. However, Australians are still more confident in the Federal government compared to during the Black Summer bushfires or just prior to the last election, but there has been a decline since the peak achieved by the Albanese government just after that election. Some of this decline appears to be due to the changing housing market, with Australians experiencing greater housing payment stress, being less satisfied with their housing situation, and there has been an increase in the level of concern about being able to ever own their own home. A key finding from the analysis is that Australians are less likely to think that home ownership matters a lot to Australia’s way of life.

January_2024_Tracking_paper.pdf ( PDF , 1.01 MB )

Tweets by ‎@WhatAustThinks

There were no tweets found.

Connect with us

Updated: 7 May 2024 / Responsible Officer: Centre Director / Page Contact: CASS Marketing & Communications

Contact ANU
Freedom of Information

+61 2 6125 5111 The Australian National University, Canberra TEQSA Provider ID: PRV12002 (Australian University) CRICOS Provider : 00120C ABN : 52 234 063 906

IMAGES

(PDF) On Database for Mobile Phones Ownership
(PDF) J2ME And Mobile Database Design
research paper on mobile apps
Mobile Phone Research Paper
Mobile Database
PPT

VIDEO

multimedia database
ICT: Chapter 14 Databases and DBMS
can we make paper mobile plan #virlshorts #techshorts
Database research paper presentation 1
paper mobile 💜 subscribe for more
Theory and implementation of database -- Research paper presentation (Dasari Vaishnavi)

COMMENTS

Mobile Data Science and Intelligent Apps: Concepts, AI-Based ...
Finally, we highlight several research issues and future directions relevant to our analysis in the area of mobile data science and intelligent apps. Overall, this paper aims to serve as a reference point and guidelines for the mobile application developers as well as the researchers in this domain, particularly from the technical point of view.
(PDF) Mobile Apps-Based Applications in Libraries and Information
A comprehensive and systematic review of the literature on mobile-based applications in libraries and information centres of peer-reviewed research papers, and conference proceedings published in ...
(PDF) A Review on Databases for Mobile Devices
Research is going on mobile database at very large. During this study different databases for mobile devices have been reviewed based on different dimensions. This paper presents and elaborates ...
Mobile Data Science and Intelligent Apps: Concepts, AI-Based Modeling
The main contributions of this paper are listed as follows: To provide a brief overview and concept of the mobile data science paradigm for the purpose of building data-driven intelligent apps. For this, we first briefly review the relevant methods and systems, to motivate our study in this area.
Mobile Databases: A Report on Open Issues and Research Directions
This paper reports on the main results of a specific action on mobile databases conducted by CNRS in France from October 2001 to December 2002. The objective was to review the state of progress in ...
PDF Mobile Database System: Role of Mobility on the Query Processing
a moving object (e.g., car of a mobile user) or from a stationary user. Consequently, the answer to a location dependent query may depend on the location of the mobile host (MH) which issued the query and/or the locations of the objects represented in the database. Therefore, an optimal query processing subsystem of a mobile database has to ...
A S ON DATA AND TRANSACTION M IN MOBILE DATABASES
This new style creates many motivating issues in mobile database research. In this paper we survey the issues related only to the data and transaction management and briefly present the state-of-the-art of data and transaction management in Mobile databases. International Journal of Database Management Systems ( IJDMS ) Vol.4, No.5, October ...
[PDF] Mobile Computing and Databases
This paper is an effort to survey these techniques and to classify this research in a few broad areas in the area of data management in mobile computing. The emergence of powerful portable computers, along with advances in wireless communication technologies, has made mobile computing a reality. Among the applications that are finding their way to the market of mobile computing-those that ...
The Research and Application of Embedded Mobile Database
Several methods to improve the data accessing rate, efficiency, stability, maintainability and uniformity of the database, including multithread technology, three-layer duplication/buffer architecture, data's optimization, and conflict detection mechanism are provided. The article discusses the characteristics of mobile and embedded database as well as its synchronization realization on ...
The Research and Application of Embedded Mobile Database
The article discusses the characteristics of mobile and embedded database as well as its synchronization realization on mobile terminal along with practical application. The article further provides several methods to improve the data accessing rate, efficiency, stability, maintainability and uniformity of the database, including multithread technology, three-layer duplication/buffer ...
PDF Issues Related To Mobile Database System
Mobile Distributed Real Time database research is still in the early stage. 4. Research overview Research community is exploring mobile database concept to great extent. A number of novel and optimized techniques has been suggested to make the concept more appropriate for the mobile world. A Dynamic Data/Currency Protocol for Mobile Database
Mobile Databases: a Report on Open Issues and Research Directions
This paper reports on the main results of a specific actionon mobile databases conducted by CNRS in France fromOctober 2001 to December 2002. The objective was toreview the state of progress in mobile databases andidentify major research directions for the French databasecommunity.
Mobile Computing and Databases-A Survey
This research has produced interesting results in areas such as data dissemination over limited bandwidth channels, location-dependent querying of data, and advanced interfaces for mobile computers. This paper is an effort to survey these techniques and to classify this research in a few broad areas.
Editorial: mobile data management and analytics
Editorial: mobile data management and analytics. This special issue of the GeoInformatica journal covers recent advances in data management and analytics in mobile, ubiquitous and pervasive computing. It contains four articles, which are extensions of the best research papers presented at the 19th IEEE International Conference on Mobile Data ...
Mobile Database System: Role of Mobility on the Query Processing
Abstract This paper reports on the main results of a specific action on mobile databases conducted by CNRS in France from October 2001 to December 2002. The objective of this action was to review the state of progress in mobile databases and identify major research directions for the French database community.
[PDF] Mobile Database Systems
2007. TLDR. This thesis presents a novel approach in selecting mobile hosts to form epidemic quorum coteries, based on metrics measured by mobile hosts and then transmitted to Base Station servers, to provide higher quorum availability and ultimately higher data accessibility, better integrity and consistency. Expand.
Data Management for Mobile Computing
Mobile Computing is a new emerging computing paradigm of the future. Data Management in this paradigm poses many challenging probl~m.~ to the database community. In this paper we iden- tify these new challenges and plan to investigate their technical significance. New research prob-
Research Trends in the Use of Machine Learning Applied in Mobile ...
This article aims to examine the research trends in the development of mobile networks from machine learning. The methodological approach starts from an analysis of 260 academic documents selected from the Scopus and Web of Science databases and is based on the parameters of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. Quantity, quality and ...
MANAGEMENT ISSUES AND CHALLENGES IN MOBILE DATABASE SYSTEM
In this paper, we survey the fundamental research challenges particular to mobile database computing, review some of the proposed solutions and identify some of the upcoming research challenges.
PDF A Survey on Mobile Databases and Query Processing in Mobile Databases
This paper describes mobile databases and query processing in mobile databases. Keywords - Base Station, Mobile Computing, Mobile Database System, Mobile Query Processing, ... research challenges, Technical Report TR-97-070, Department of Computer Sciences, Purdue University, 1993.
Mobile learning: research context, methodologies and future works
For the qualitative research method, data was collected via task analysis and interviews. Data were captured through multiple channels including video data analysis and interview content analysis. ... The country with the highest number of research papers in the field of mobile learning is China with 11 articles. There is a lack of study in ...
Unveiling large-scale commuting patterns based on mobile phone cellular
Abstract. In this study, with Estonia as an example,we established an approach based on Hidden Markov Model to extract large-scale commuting patterns at different geographical levels using a massive amount of mobile phone cellular network data, which is referred to as Call detail record (CDR). The proposed model is designed for reconstructing ...
Advances in database systems education: Methods, tools, curricula, and
In order to preserve the principal aim of this study, which is to review the research conducted in the area of database systems education, a piece of advice has been collected from existing methods described in various studies (Elberzhager et al., 2012; Keele et al., 2007; Mushtaq et al., 2017) to search for the relevant papers. Thus, proper ...
Why Tabular Foundation Models Should Be a Research Priority
Recent text and image foundation models are incredibly impressive, and these models are attracting an ever-increasing portion of research resources. In this position piece we aim to shift the ML research community's priorities ever so slightly to a different modality: tabular data. Tabular data is the dominant modality in many fields, yet it is given hardly any research attention and ...
10 Data Science Papers for Academic Research in 2024
Data science is a rapidly evolving field, with new research and advancements emerging every year. As we approach 2024, academic researchers must stay up-to-date with the latest developments in data science. In this article, we present a curated list of ten data science papers that are expected to be influential in the coming year. These papers cover a wide range of topics, including machine ...
Applied Sciences
SLAM (Simultaneous Localization and Mapping), primarily relying on camera or LiDAR (Light Detection and Ranging) sensors, plays a crucial role in robotics for localization and environmental reconstruction. This paper assesses the performance of two leading methods, namely ORB-SLAM3 and SC-LeGO-LOAM, focusing on localization and mapping in both indoor and outdoor environments.
Nightside clouds and disequilibrium chemistry on the hot ...
From our Eureka! v1 analysis (), we measure a broadband (5-10.5 μm) peak-to-trough phase variation of 4,180 ± 33 ppm with an eclipse depth of 5,752 ± 19 ppm and a nightside flux of 1,636 ± ...
Global Smartphone Market Grows 6% YoY in Q1 2024; Revenue Reaches
Commenting on Apple's performance, Research Director Jeff Fieldhack said, "Tough competition in China, record low upgrades in the US and a difficult compare from last year due to iPhone 14 Pro's supply shifting to Q1 2023 all weighed on iPhone performance. However, there were upsides as well. An improved product mix with 15 Pro's performing better than its predecessors, and an ...
[2405.00208] A Primer on the Inner Workings of Transformer-based
The rapid progress of research aimed at interpreting the inner workings of advanced language models has highlighted a need for contextualizing the insights gained from years of work in this area. This primer provides a concise technical introduction to the current techniques used to interpret the inner workings of Transformer-based language models, focusing on the generative decoder-only ...
It's the economy (and housing), stupid: Views of Australians on the
This paper summarises recent public opinion data on views on the economy, as well as experience and views on the housing market. It is based on data from the January 2024 ANUpoll as well as historic ANUpoll data to show that compared to almost any time during the pandemic or immediately beforehand, Australians are more financially stressed. They are also less satisfied with the direction of ...

The Research and Application of Embedded Mobile Database

Purchase Details

Profile Information

Mobile Database System: Role of Mobility on the Query Processing

Related Papers

RELATED PAPERS

Mobile learning: research context, methodologies and future works towards middle-aged adults – a systematic literature review

Cite this article

1 Introduction

2 Mobile learning

3 Mobile technology

4 Multimedia in Mobile learning application

5 Mobile learning theory

6 Methodology

7 Results of the systematic mapping

8 Participants

9 Research context in Mobile learning

10 Mobile learning towards the middle-aged adults

11 Research methodology

12 Discussion

13 Conclusion and future work

Acknowledgments

Conflict of interest

Availability of data and materials

Author information

Corresponding author

Additional information

Rights and permissions

About this article

Share this article

Nightside clouds and disequilibrium chemistry on the hot Jupiter WASP-43b

Similar content being viewed by others

Diurnal variations in the stratosphere of the ultrahot giant exoplanet WASP-121b

A broadband thermal emission spectrum of the ultra-hot Jupiter WASP-18b

UV absorption by silicate cloud precursors in ultra-hot Jupiter WASP-178b

Model interpretation

Observations and quality of the data

Data reduction pipelines

Eureka! v2 reduction

TEATRO reduction

SPARTA reduction

Spectral binning

Light-curve fitting

Shadowed region effect

Determining the number of sinusoid harmonics

Eureka! v1 fitting methods

Eureka! v2 fitting methods

TEATRO fitting methods

SPARTA fitting methods

Combining independent spectra

Atmospheric forward models

Generic PCM

SPARC/MITgcm with radiative transfer post-processing by gCMCRT

expeRT/MITgcm

Atmospheric retrieval models

HyDRA retrieval framework

PyratBay retrieval framework

NEMESIS retrieval framework

SCARLET retrieval framework

PLATON retrieval framework

ARCiS retrieval framework

Data availability

Code availability

Acknowledgements

Author information

Contributions

Corresponding author

Ethics declarations

Peer review

Additional information

Extended data

Extended Data Fig. 2 A model-independent demonstration of the initial changes in flux for the WASP-43b observations.

Extended Data Fig. 3 Retrieved spectra from the six retrievals.

Extended Data Fig. 4 Chemically-consistent atmospheric retrievals.

Extended Data Fig. 5 Retrieval contribution functions.

Extended Data Fig. 6 PyratBay clouds exploration.

Extended Data Fig. 7 A comparison of the retrieved temperature-pressure profiles to the GCM simulations.

Supplementary information

Rights and permissions

About this article