Machine learning model developed by Amigo s.r.l. for the SnowPower solution, part of the I-NERGY 1st Open Call. 

The model reads a dataset of features and returns a daily SWE indicator estimates for several Alpine basins (and is further described in https://www.ai4europe.eu/research/ai-catalog/description-and-setup-ml-models-estimation-snow-water-equivalent-swe-and-runoff). The resulting SWE indicator is provided over the watersheds visible in https://github.com/amigoclimateteam/SnowPower/blob/main/SnowPower_SWE_data_2017_2020_Asset3/basins_mapping.PNG.

The model was trained on fine-scale SWE data that were previously averaged over each watershed, standardized, and finally shifted in order to avoid negative values. For this reason we refer to the results obtained through this model as SWE indicator, as they maintain the seasonality and spatial patterns that are typical of SWE, albeit without dimensional units. This data can thus be used as input for machine learning and deep learning models.