With the rapid growth of remote sensing imaging technology, vast amounts of remote sensing data are being generated, which presents nontrivial challenges for monitoring Earth, Mars, and other planets, as well as for applications in national security, agriculture, and atmospheric studies, to name a few. In recent decades, deep learning techniques have had significant impact on multi-source data processing and analysis, particularly in classification, super-resolution, and detection, among other areas. However, several challenges remain due to the limited availability of annotated datasets, restricted computing resources, the complexity and diversity of large-scale areas, and other specific issues, all of which make deep learning-based algorithms more difficult in real-world applications. Therefore, novel deep neural networks, combined with few-shot learning, meta-learning, attention mechanism, or other emerging transformer/mamba technologies, require more attention. This is of vital importance in multi-source data processing and applications, where data heterogeneity and complexity pose significant challenges. Moreover, it is crucial to develop lightweight, explainable, and robust networks for multi-source data processing.

This special section aims to explore deep networks for more accurate and efficient multi-source data processing. In particular, it seeks to achieve cross-domain performance with high efficiency through lightweight network design. Authors are encouraged to submit research articles, review articles, or application-oriented articles on topics related to multi-source data classification, semantic segmentation, detection, spectral super-resolution, and understanding. These topics include but are not limited to the following:

• Machine and deep learning-based algorithms for multi-source data processing
• Multi-source image processing and pattern recognition
• Joint classification and semantic segmentation
• Spectral super-resolution
• Target detection and change detection
• Lightweight deep neural networks
• Domain adaptation, few-shot learning, and meta-learning-based algorithms
• Onboard real-time applications using multi-source datasets

To submit a manuscript for consideration, please prepare the manuscript according to the journal guidelines and submit the paper via the online submission system ( https://jars.msubmit.net). A cover letter indicating that the submission is intended for this special section should be included with the paper. Papers will be peer reviewed in accordance with the journal’s established policies and procedures. Peer review will commence immediately upon manuscript submission, with a goal of making a first decision within six weeks of manuscript submission. The special section is opened online once a minimum of four papers have been accepted. Each paper is published as soon as the copyedited and typeset proofs are approved by the author.

The small data problem is a significant challenge in remote sensing (RS) data analysis. Despite the vast availability of RS imagery, acquiring sufficient ground-truth response labels to train predictive models remains a difficult and often resource-intensive task. This challenge arises because most applications in remote sensing aim to develop models that can accurately predict specific environmental variables or phenomena based on various observed input features. Traditional deep learning models are highly data-driven, often requiring large amounts of labeled data to learn complex patterns and deliver reliable predictions. When such data is scarce, even the most sophisticated models fall short of realizing the full potential of remote sensing imagery. This not only impacts model accuracy and reliability but also slows the pace of innovation, limiting the real-world applications of remote sensing in areas like environmental monitoring, disaster response, and climate change studies. To bridge this gap, it is essential to develop new strategies that maximize the value of available data, offering more effective and practical solutions to pressing global challenges.

Recent advancements, including data augmentation, domain adaptation, transfer learning, and other novel approaches, aim to overcome these limitations. The Journal of Applied Remote Sensing invites researchers to contribute to this special section and invite high-quality original research articles, case studies, and comprehensive reviews that offer new perspectives and solutions in this critical area of remote sensing research.

Topics of interest include but are not limited to:

• Data augmentation methods for remote sensing applications
• Domain adaptation approaches to bridge the gap between source and target domains
• Transfer learning and few-shot learning for remote sensing image classification and segmentation
• Synthetic data generation techniques for enhancing limited training datasets
• Self-supervised and unsupervised learning approaches for remote sensing data
• Case studies and applications demonstrating the use of limited training samples in real-world remote sensing scenarios
• Cross-domain adaptation and multi-source data fusion strategies for robust learning
  

To submit a manuscript for consideration, please prepare the manuscript according to the journal guidelines and submit the paper via the online submission system ( https://jars.msubmit.net). A cover letter indicating that the submission is intended for this special section should be included with the paper. Papers will be peer reviewed in accordance with the journal’s established policies and procedures. Peer review will commence immediately upon manuscript submission, with a goal of making a first decision within six weeks of manuscript submission. The special section is opened online once a minimum of four papers have been accepted. Each paper is published as soon as the copyedited and typeset proofs are approved by the author.

Oceans cover more than 70% of the Earth’s surface and provide over 50% of the world’s oxygen and store carbon dioxide. In addition, oceans transport heat from the equator for the poles and regulate climate patterns. Therefore, investigating ocean activities and comprehending the role of the oceans in global climate change is essential. However, existing ocean-observation data have deficiencies, such as inconsistent spatial distribution and restricted observation depth layers. Combining multi-source observation data (in situ observation data, sea surface data, numerical model data, and reanalysis data) can effectively predict and forecast the spatial-temporal evolution of ocean meteorology and marine environments.

This special section will focus on fusing multi-source ocean remote sensing data to address the challenges posed by the cross-modal features correlation and remote sensing big data. The aim is to solve the critical problems of artificial intelligence in oceanography. The Journal of Applied Remote Sensing invites researchers to contribute to this section, showcasing the latest advances and the applications of ocean remote sensing data.

Topics of interest include but are not limited to the following:

• Fusion techniques for multi-source ocean observation and remote sensing data.
• Physics-guided ocean data prediction and forecasting.
• Predicting and forecasting typical ocean disasters based on multi-source observation and remote sensing data.
• Cognition of ocean processes based on multi-source observation and remote sensing data.
• Ocean color observation and applications.
• Ocean internal wave observation and applications.
• Spectrum recovery from multispectral ocean color data.
• Coastal wetland mapping and monitoring.
• Deep learning techniques for ship performance monitoring.
• Modeling and change analysis of the Arctic.
• Oil spill detection on the ocean via multi-source data.

To submit a manuscript for consideration, please prepare the manuscript according to the journal guidelines and submit the paper via the online submission system ( https://jars.msubmit.net). A cover letter indicating that the submission is intended for this special section should be included with the paper. Papers will be peer reviewed in accordance with the journal’s established policies and procedures. Peer review will commence immediately upon manuscript submission, with a goal of making a first decision within six weeks of manuscript submission. The special section is opened online once a minimum of four papers have been accepted. Each paper is published as soon as the copyedited and typeset proofs are approved by the author.