Advanced computer models are increasingly used to predict weather and climate. While powerful, these models often face challenges such as becoming unstable over long periods or being too expensive to run at high detail. Researchers recently developed LUCIE, a climate model that accurately predicts climate trends for up to 100 years using fewer computing resources. However, LUCIE provides forecasts at a broad scale, which are not detailed enough for local climate assessments needed by communities and policymakers.

In our study, we created a new approach to transform LUCIE’s large-scale predictions into detailed local forecasts. We used advanced artificial intelligence techniques that take coarse climate data and produce high-resolution outputs suitable for regional climate impact studies. We tested our models rigorously using real-world data and compared their performance using various quality measures. Our results show that this method accurately preserves large-scale climate features while providing detailed local information, offering a valuable tool for climate planning and adaptation efforts.