Weather Forecasting, Reanalysis, Downscaling

There is an increasing interest in the application of Deep Leaning Techniques for modeling weather dynamics and investigating climatic changes. This cover various different domains, covering in particular

• Reanalys: the goal is to produce a “best possible” record for climate analysis by filtering, combining and integrating historical observations. The output is supposed to be a consistent, regularly gridded, high-resolution estimate of the atmospheric state over decades.
• Forecasting: the goal is to predict futore weather conditions from hours to weeks ahead, with uncertainty growing over time. The research consists in simulating or integrating dynamical models with data-driven neural systems.
• Downscaling: this task aims to bridge the gap between coarse-resolution model outputs (from reanalysis or forecasts) and local-scale phenomena.

---

• Andrea Asperti, Ali Aydogdu, Angelo Greco, Fabio Merizzi, Pietro Miraglio, Beniamino Tartufoli, Alessandro Testa, Nadia Pinardi, Paolo Oddo. Deep Learning for Sea Surface Temperature Reconstruction under Cloud Occlusion

Generative models represent the state-of-the-art approach for addressing problems characterized by a substantial degree of stochasticity in the predicted outcomes. In essence, the focus lies in modeling the probability distribution of the results, wherein the expected value signifies the most probable prediction. The precise goal of generative models is to accurately capture this distribution.

This approach finds applications across a wide array of disciplines, encompassing domains such as weather forecasting, financial analysis and prediction, epidemiology and disease spread, traffic flow and transportation modeling, as well as social dynamics and opinion propagation.

• Asperti, A., Merizzi, F., Paparella, A. et al. Precipitation nowcasting with generative diffusion models. Appl Intell 55, 187 (2025). https://doi.org/10.1007/s10489-024-06048-y

The Copernicus Regional Reanalysis for Europe, CERRA, is a high-resolution regional reanalysis dataset for the European domain. In recent years, it has shown significant utility across various climate-related tasks, ranging from forecasting and climate change research to renewable energy prediction, resource management, air quality risk assessment, and the forecasting of rare events, among others. Unfortunately, the availability of CERRA is lagging 2 years behind the current date, due to constraints in acquiring the requisite external data and the intensive computational demands inherent in its generation. As a solution, this paper introduces a novel method using diffusion models to approximate CERRA downscaling in a data-driven manner, without additional informations. By leveraging the lower resolution ERA5 dataset, which provides boundary conditions for CERRA, we approach this as a super-resolution task. Focusing on wind speed around Italy, our model, trained on existing CERRA data, shows promising results, closely mirroring the original CERRA. Validation with in-situ observations further confirms the model’s accuracy in approximating ground measurements.

• F.Merizzi, A.Asperti, S.Colamonaco. Wind speed super-resolution and validation: from ERA5 to CERRA via diffusion models. Neural Comput & Applic 36, 21899–21921 (2024). https://doi.org/10.1007/s00521-024-10139-9