arXiv:2504.00307v2 Announce Type: replace 
Abstract: Improving the representation of precipitation in Earth system models (ESMs) is critical for assessing the impacts of climate change and especially of extreme events like floods and droughts. In existing ESMs, precipitation is not resolved explicitly, but represented by parameterizations. These typically rely on resolving approximated but computationally expensive column-based physics, not accounting for interactions between locations. They struggle to capture fine-scale precipitation processes and introduce significant biases. We present a novel approach, based on generative machine learning, which integrates a conditional diffusion model with a UNet architecture to generate accurate, high-resolution (0.25{\deg}) global daily precipitation fields from a small set of prognostic atmospheric variables. Unlike traditional parameterizations, our framework efficiently produces ensemble predictions, capturing uncertainties in precipitation, and does not require fine-tuning by hand. We train our model on the ERA5 reanalysis and present a method that allows us to apply it to unseen ESM data, enabling fast generation of probabilistic forecasts and climate scenarios. By leveraging interactions between global prognostic variables, our approach provides an alternative parameterization scheme that mitigates biases present in the ESM precipitation while maintaining consistency with its large-scale (annual) trends. This work demonstrates that complex precipitation patterns can be learned directly from large-scale atmospheric variables, offering a computationally efficient method to obtain high-resolution precipitation without the cost of running the dynamical model at such high resolution.

تم تقديم نهج جديد يستخدم التعلم الآلي التوليدي لتحسين تمثيل حقول هطول الأمطار العالمية في نماذج النظام الأرضي (ESMs). يدمج هذا الأسلوب نموذج انتشار شرطي مع بنية UNet لتوليد بيانات هطول الأمطار اليومية عالية الدقة من مجموعة محدودة من المتغيرات الجوية، مما يعالج التحيزات الموجودة في المعلمات التقليدية.

Se ha introducido un enfoque novedoso que utiliza el aprendizaje automático generativo para mejorar la representación de los campos de precipitación global en los modelos de sistema terrestre (ESM). Este método integra un modelo de difusión condicional con una arquitectura UNet para generar datos de precipitación diaria de alta resolución a partir de un conjunto limitado de variables atmosféricas, abordando los sesgos presentes en las parametrizaciones tradicionales.

Une nouvelle approche utilisant l'apprentissage machine génératif a été introduite pour améliorer la représentation des champs de précipitations mondiales dans les modèles de système terrestre (ESM). Cette méthode intègre un modèle de diffusion conditionnelle avec une architecture UNet pour générer des données de précipitations quotidiennes à haute résolution à partir de variables atmosphériques limitées, abordant ainsi les biais présents dans les paramétrisations traditionnelles.

A novel approach utilizing generative machine learning has been introduced to enhance the representation of global precipitation fields in Earth system models (ESMs). This method integrates a conditional diffusion model with a UNet architecture to generate high-resolution daily precipitation data from limited atmospheric variables, addressing the biases present in traditional parameterizations.