Inside the Journal Where AI Meets Weather, Climate and Oceans

Most people will never read a peer-reviewed journal article about machine learning and atmospheric science. But the research happening in these pages shapes the weather app on your phone, the flood warnings your county issues, and the climate projections that inform how cities plan for the next fifty years. That's the quiet significance of Artificial Intelligence for the Earth Systems, known in research circles as AIES.

Published by the American Meteorological Society, AIES is a fully open-access journal, meaning anyone can read the research without a subscription paywall. Since its launch in 2022, it has become a hub for scientists exploring how AI and machine learning can improve our understanding of weather, climate, oceans, and hydrology. Amy McGovern of the University of Oklahoma serves as editor in chief, overseeing a field that is expanding fast. The journal now carries an impact factor of 4.5, a metric used to gauge how often research gets cited by other scientists, placing it among the more influential outlets in both meteorology and artificial intelligence research categories.

Think of AIES as a translation hub. On one side sit traditional climate and weather models, the kind built from decades of physics equations describing how heat, moisture, and air pressure interact. On the other side sits machine learning, a set of techniques that let computers find patterns in massive datasets without being explicitly programmed for every scenario. The scientists publishing in AIES are trying to combine the two: using AI not to replace physical understanding, but to sharpen it, speed it up, or fill in gaps where traditional models struggle.

What the Research Actually Covers

The scope is broad by design. Recent papers touch on tornado detection, drought prediction, aerosol chemistry, seasonal forecasting, and even how radar systems can better track severe storms. One paper from Ryan Martz and colleagues examines detection methods relevant to severe weather warnings. Another, from Zhehao Liang, Paola Crippa, and Stefano Castruccio, likely wades into questions of climate variability. A team led by Bhupendra Raut published a perspectives piece drawing on work at Argonne National Laboratory, where researchers have been testing AI-powered observation systems in the field.

Some of the journal's most cited work since 2022 offers a window into where the field's biggest breakthroughs have landed. A 2023 paper by Lily-belle Sweet, Christoph Müller, Mohit Anand, and Jakob Zscheischler has been cited 46 times, the most of any AIES article in that window. Close behind is a 2022 paper by Antonios Mamalakis, Elizabeth Barnes, and Imme Ebert-Uphoff, cited 45 times, which likely deals with explainability, a growing concern in AI research broadly. When a machine learning model predicts a hurricane's path or a heat wave's severity, forecasters and the public alike want to know why the model reached that conclusion, not just what it concluded. Explainable AI tries to answer that question, opening up the "black box" so scientists can trust, and correct, the systems they build.

Other highly cited work addresses downscaling, a technique for taking coarse, low-resolution climate model output and refining it into detail useful at the local level. A 2024 paper by Neelesh Rampal and eight coauthors, cited 45 times, tackles exactly this problem. Downscaling matters because a global climate model might tell you how the planet warms on average, but a city planner needs to know how that translates to flooding risk on a specific street. AI is increasingly the tool bridging that gap.

The journal also publishes foundational overview pieces. A 2022 paper from Peter Dueben, Martin Schultz, Matthew Chantry, David John Gagne II, David Matthew Hall, and Amy McGovern, cited 29 times, appears to have helped set the research agenda for the field's early years. Papers like this often function as a map, showing newer researchers where the unexplored territory lies.

Why It Matters

There's a real tension running through all of this work, and it's worth sitting with. Machine learning models can be remarkably good at pattern recognition, sometimes outperforming traditional physics-based forecasts on narrow tasks like short-term rainfall prediction. But they can also fail in unpredictable ways when faced with conditions outside their training data, which is a genuine risk when the stakes involve evacuation orders or agricultural planning. The journal's inclusion of "ethical and responsible use of AI/ML" as an explicit topic area isn't a footnote. It reflects a field grappling honestly with its own limitations.

There's also a quieter benefit to open access publishing here that deserves attention. Climate and weather AI research often depends on massive datasets and computing resources that wealthier institutions and countries can access more easily. By making this journal free to read, the American Meteorological Society lowers one barrier, letting researchers in resource-limited settings, students, and even curious members of the public engage with cutting-edge climate science without a subscription fee standing in the way.

None of this happens in isolation from the rest of atmospheric science. AIES sits alongside other AMS publications like the Bulletin of the American Meteorological Society and draws on decades of institutional knowledge, including the society's Glossary of Meteorology and its ongoing State of the Climate reporting. The AI tools being developed and tested in these pages will increasingly inform how that broader body of climate knowledge gets built and communicated in the years ahead.

For a field moving as quickly as AI-driven climate science, having a dedicated, peer-reviewed, freely accessible venue matters more than it might seem at first glance. It's where the guardrails get built, the mistakes get caught before they reach operational forecasting systems, and the next generation of tools gets vetted. That work rarely makes headlines. But when a flood warning arrives an hour earlier than it would have a decade ago, or a seasonal drought forecast proves accurate enough for farmers to plan around, this is often where the underlying science started.