AI for Good in Climate Science: Real-World Machine Learning and Deep Learning for Climate Monitoring, Adaptation, and Mitigation

Climate science increasingly relies on large, heterogeneous, and often imperfect datasets, including satellite observations, in situ measurements, reanalysis products, model outputs, and hybrid environmental data. Machine learning and deep learning offer powerful opportunities to extract patterns, integrate multiple data streams, generate predictions, and support decision-relevant analysis across climate-related domains. However, many AI-oriented studies still suffer from limited transparency, weak reproducibility, incomplete methodological reporting, restricted data accessibility, and uncertain real-world relevance. This Research Topic addresses this gap by promoting climate-focused applications of AI that combine methodological rigor with interpretability, openness, reproducibility, and practical value, while explicitly linking computational advances to climate monitoring, adaptation, mitigation, risk, and Earth system understanding.This Research Topic aims to demonstrate how machine learning and deep learning can support climate science in meaningful, transparent, and actionable ways. The goal is not to showcase artificial intelligence as an end in itself, but to identify approaches that clearly improve climate monitoring, strengthen adaptation strategies, support mitigation efforts, or advance understanding of climate variability and change. We seek contributions that connect data, methods, interpretation, and real-world value under realistic conditions, with explicit treatment of validation, uncertainty, limitations, and applicability. Particular emphasis will be placed on reproducible and FAIR research practices, including clear data availability statements and, whenever possible, open datasets, code, trained models, and computational workflows. By bringing together rigorous applications across atmosphere, ocean, land, and polar systems, the collection will help distinguish scientifically useful AI from purely technical or benchmark-driven advances.We welcome original research, reviews, methods-oriented studies, perspectives, and interdisciplinary contributions with explicit climate relevance. Themes include ML/DL for climate and environmental monitoring; remote sensing; extreme-event detection and prediction; climate risk and hazard assessment; adaptation and resilience; mitigation and emissions analysis; physics-informed and hybrid data-driven frameworks; spatiotemporal modeling; downscaling; emulation; anomaly detection; and climate decision-support systems. Contributions should clearly address transparency, validation, uncertainty, reproducibility, and realistic applicability. Broader environmental applications are welcome only when their connection to climate change, climate risk, adaptation, or mitigation is explicit. We particularly encourage balanced coverage across atmosphere, ocean, and land, including polar systems, long-term climate variability and trends, and studies that make data, code, models, or workflows openly accessible whenever possible. Topic editors: Andre L. Belem, Imee Necesito, Jonas Takeo Carvalho, Jia Zhang Source: https://www.frontiersin.org/research-topics/85560/ai-for-good-in-climate-science-real-world-machine-learning-and-deep-learning-for-climate-monitoring-adaptation-and-mitigation