Author(s): Payam Heidarian; Matteo Benetti; Marco Pilotti; Marco Gabella; Esmail Ghaemi
Linked Author(s): Marco Pilotti
Keywords: Remote Sensing ICON Model Ground-based station data weather Forecasting Machine learning
Abstract: The use of weather forecasting models and precipitation forecasts derived from radar data has become a crucial tool for flood forecasting in hydrological applications. These models, which will rely heavily on real-time radar data and forecasted radar-based information, provide high temporal and spatial resolution. However, their accuracy can sometimes be limited by discrepancies in radar estimates. This study will aim to improve precipitation forecasts by enhancing radar-based predictions with ground station data from the Franciacorta basin, located in northern Italy. The main objective of this work will be to address the gap between radar-based forecasts and ground station measurements, thus improving the predicted precipitation estimates through the application of a machine learning algorithm. To achieve this, different precipitation events will be selected for training the machine learning model, with another 5 events will be used for testing. The results of this research will highlight the effectiveness of using machine learning techniques to adjust precipitation forecast data, bringing it closer to ground station measurements and improving overall accuracy in flood forecasting.
Year: 2025