S3M Italy: a real-time, open-source cryospheric-forecasting chain for applications on a large scale

Monitoring the state of the cryosphere in real time is a key to improved risk and water resources management, especially in a warming climate. All around the world, this goal is achieved through forecasting chains combining models with in-situ and remote-sensing measurements. Here, we discuss lessons learned while developing S3M Italy, one such chain delivering hourly estimates of snow water equivalent, density, snow and glacier melt, and bulk liquid water content across the Italian territory (300k+ km2, 200 m resolution, 1.5 hour turnaround). S3M Italy includes downloaders to ingest input data from automatic weather stations, spatialization tools to convert these data into weather-input maps, blending routines for deriving daily snow-covered-area maps from ESA Sentinel 2, NASA MODIS, and EUMETSAT H-SAF products, mapping algorithms based on multilinear regressions for assimilating on-the-ground snow-depth data, as well as algorithms to manage parallelized runs and then mosaic model outputs. S3M Italy has been developed to support decisions by the Italian Civil Protection Agency and is fully open source, not only in terms of underlying models (https://github.com/c-hydro/s3m-dev), but also in terms of all pre-processing routines (https://github.com/c-hydro/fp-hyde, https://github.com/c-hydro/fp-s3m).