A Comprehensive Snow Modeling Using Multi-Source Data and Assimilation for a Refined Characterization of a Complex Mediterranean Basin

Climate change significantly impacts snow dynamics, thereby affecting water resources, especially in countries like Türkiye, where snow is crucial for water supply. This study focuses on one of the largest Mediterrenain river basins, the Seyhan Basin (21,890 km²), contributing significantly to Türkiye's overall water resources. The basin's extensive agricultural activities and significant hydropower potential necessitate sustainable water management for its long-term sustainability. The basin's complex mountainous topography and its location at the intersection of Mediterranean and continental climate zones, combined with limited data availability, present significant challenges for hydrological modeling. These factors make the Seyhan Basin an ideal region for analyzing changes in snow potential and related water resources.

This study aims to refine spatial snow characterization in this complex Mediterranean basin through comprehensive snow modeling using multi-source data and assimilation. The spatial and temporal accuracy and reliability of Snow Multidata Mapping and Modeling (S3M) model outputs for snow-water equivalent, snow depth, and snow-covered area are assessed. The main S3M model inputs, derived from ERA5-Land, include hourly temperature, relative humidity, shortwave radiation, and precipitation data from 2012 to 2022. Model inputs of temperature and precipitation are validated against observations from 12 meteorological stations within and around the basin. The S3M data assimilation framework improves model estimates by incorporating satellite snow cover area data (Eumetsat H SAF products). Additionally, snow-covered area estimates are compared to MODIS, IMS, and ERA5-Land datasets, while snow-water equivalent measurements from five in situ stations, ERA5-Land and H SAF datasets provide independent validation for SWE outputs. The performance of daily aggregated model results is evaluated using different metrics as RMSE, KGE, and NSE, besides spatial performance analysis as false alarm rate and hit scores for the whole period. The results indicate that NSE performance is 0.90-0.95 for SCA, RMSE is 5-30 mm for SWE and the false alarm rate is calculated as 0.15-0.35 for SCA.