Spatiotemporal AI Methods for Atmospheric Remote Sensing

Dear Colleagues,

Spatiotemporal Artificial Intelligence (AI) methods are revolutionizing atmospheric remote sensing by enabling a more accurate, efficient, and scalable analysis of the Earth’s atmosphere. Atmospheric constituents and phenomena, such as aerosols, clouds, greenhouse gases, and extreme weather, exhibit highly dynamic behavior across space and time. Capturing their variability is essential for improving global and regional weather prediction, air quality monitoring, climate change assessment, and disaster risk mitigation.

However, traditional retrieval and analysis methods face significant challenges when dealing with atmospheric data, particularly from remote sensing platforms, including the following:

• Massive Data Volume: Modern satellite constellations generate petabytes of data at high spatial, spectral, and temporal resolutions, often exceeding the capacity of conventional processing pipelines.
• Data Heterogeneity: Observations originate from diverse instruments (e.g., UV/Vis, IR, and microwave), platforms (satellites, aircraft, and ground stations), and formats, complicating integration and interpretation.
• Non-Linearity and Complexity: Atmospheric processes involve highly nonlinear interactions influenced by cloud microphysics, radiative transfer, and boundary-layer dynamics, making traditional physics-based models insufficient for some tasks.

Spatiotemporal AI addresses these challenges through advanced techniques in machine learning, deep learning, computer vision, and time-series modeling, offering novel capabilities such as the following:

• Multimodal data fusion across instruments and platforms;
• Scene-dependent feature extraction that adapts to variability in geography and viewing geometry;
• Anomaly detection for identifying outliers, artifacts, or regime shifts in environmental observations;
• Forecasting and nowcasting using recurrent neural networks, transformers, and spatiotemporal graph models;
• Uncertainty quantification via Bayesian deep learning or ensemble approaches.

By integrating AI-driven methodologies with remote sensing data, researchers can extract deeper insights from continuous global observations including the following:

• Cloud property retrieval (e.g., optical depth, phase, and height) using deep convolutional neural networks;
• Aerosol and trace gas estimation through hybrid physical–statistical inversion models;
• Atmospheric motion vector extraction for improving numerical weather prediction;
• Pollution hotspot forecasting to support public health and urban planning;
• Detection of extreme weather events, such as hurricanes, wildfires, and dust storms;
• Long-term trend analysis to evaluate climate impacts and monitor emissions.

This Special Issue sits at the intersection of atmospheric science, artificial intelligence, remote sensing, environmental health, and data science, offering a truly interdisciplinary approach. It not only pushes the frontier of methodological innovation but also delivers operational value by supporting critical societal needs in hazard forecasting, environmental management, and sustainable development.

As the volume and diversity of remote sensing data from next-generation satellite constellations, aircraft campaigns, and ground networks continue to grow, there is an urgent need for AI-driven solutions that can process, interpret, and apply this data at different scales. Thus, the combination between spatiotemporal AI and remote sensing will become indispensable, and fostering this integration will accelerate both scientific discovery and practical applications in environmental monitoring, making Earth observation systems smarter, more adaptive, and more responsive to the needs of humanity.

This Special Issue invites high-quality research contributions that advance innovative spatiotemporal AI methodologies for atmospheric remote sensing, with a focus on capturing the complex and dynamic phenomena of Earth’s atmosphere. In addition to technical advancements, we aim to showcase research with meaningful societal impact, particularly in areas such as climate action and monitoring, public health and environmental justice, disaster preparedness and resilience, and sustainable development and urban planning. By emphasizing both methodological innovation and real-world applications, this Special Issue strives to bridge the gap between cutting-edge AI techniques and their practical value for atmospheric science and broader societal benefit.

We welcome interdisciplinary studies that integrate AI, remote sensing, atmospheric modeling, computer vision, GIS, and environmental sciences to better understand spatial and temporal patterns in atmospheric composition and behavior. Relevant topics include, but are not limited to, the following:

• Deep learning and spatiotemporal AI models for analyzing atmospheric factors and compositions using remote sensing.
• Data fusion and harmonization of multi-platform, multi-resolution, and multi-spectral remote sensing observations.
• AI-enhanced retrieval algorithms for aerosols, clouds, trace gases, and temperature profiles.
• Novel techniques for wildfire smoke detection, transport modeling, and impact evaluation.
• Analysis of particulate matter (PM2.5/PM10), dust storms, haze episodes, and their regional transport mechanisms.
• Integration of remote sensing, GIS, and AI for hazard mapping, pattern recognition, and spatial epidemiology.
• Forecasting of air quality, pollutant transport, and atmospheric visibility using machine learning and hybrid models.
• Uncertainty quantification, explainability, and interpretability of AI models in atmospheric applications.
• Spatiotemporal downscaling, super-resolution, and gap-filling techniques for improving remote sensing data utility.
• Use of benchmark datasets, synthetic data generation, and open-source AI models for reproducible research.
• Next-generation satellite mission data processing technologies, including onboard AI, edge computing, and real-time analytics.
• Spatiotemporal downscaling and super-resolution techniques for remote sensing datasets.

Dr. Qian Liu
Dr. Manzhu Yu
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Remote Sensing is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• spatiotemporal AI
• remote sensing
• atmospheric composition
• cloud features
• air pollution and aerosols
• environmental monitoring
• climate and health impacts