Original title: A Checklist to assess the energy and carbon impacts of ML/AI applications in Earth System Modeling
Filippo Dainelli, Amirpasha Mozaffari, Marina Castaño, Aina Gaya i Àvila +4 · September 2026
Scientists increasingly use machine learning and artificial intelligence to model weather, climate, and Earth systems. However, building these computer models requires computing power that consumes energy and produces carbon emissions. Many guidelines for eco-friendly computing exist, but they are scattered across different papers, making them hard to use in daily research. To solve this problem, the researchers gathered these scattered ideas into a practical checklist. The checklist follows the step-by-step pipeline of model development. Researchers can use it to evaluate and lower the environmental impact of their projects. The authors also gathered specific metricsStandard measurements used to estimate or compare something, here energy use and emissions. from past studies to estimate energy use and carbon footprints, offering clear examples and actionable tips to help scientists make greener choices during everyday work.
You are cooking a big meal for guests. You have a limited amount of gas and electricity, and you want the food to be great without wasting energy. But there are so many small choices—which burner, how hot, whether to preheat—that it is easy to use far more energy than you noticed. In this paper, the meal is the computer program being built, and the checklist makes the people building it pause at each step and ask whether they really need that much energy. It helps them finish the same project while wasting less.
Researchers building machine learning into climate and weather models now have a practical way to check its environmental cost. The checklist turns scattered advice into concrete questions at each stage of development, so labs can reduce energy and carbon without giving up the science.
The abstract doesn't report any measured energy or carbon reductions from using the checklist, nor does it introduce new metrics. It compiles existing advice and metrics, so the checklist's real-world effect on researchers' behavior remains untested here.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
A Checklist to assess the energy and carbon impacts of ML/AI applications in Earth System Modeling — Filippo Dainelli, Amirpasha Mozaffari, Marina Castaño et al.
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.