Assimilation of ERA5-land and SNODAS SWE Products to Enhance Spring Flow Predictions in Southern Quebec, Canada

While streamflow observation has been considered and proved its importance in hydrological data assimilation in many studies, snow-related data, such as SWE, also contains vital information regarding water volume in the spring season. Therefore, assimilating this data into a hydrological model is expected to ameliorate the simulated snow-related model states, such as SWE, snow depth, and snow density, and accordingly, enhance spring flow predictions in snow-dominated regions. However, point-scale in-situ SWE data are rarely available in many areas, including Quebec, Canada. Moreover, using interpolation methods to generate observed data from point measurements introduces additional uncertainty and subsequent challenges in quantifying the related errors. As a result, spatially distributed SWE data for the purpose of this study was extracted from two different datasets, namely ERA5-land reanalysis and SNOw Data Assimilation System (SNODAS). ECMWFs ERA5-land dataset that spans from January 1950 to near real time is a replay of the land component of the ERA5 climate reanalysis with some improvements that enhances its resolution from 31 km in ERA5 to 9 km in ERA5-land. On the other hand, NOHRSCs SNODAS is a snow data assimilation system that improves outputs of a snow model by assimilating observed snow data provided by airborne platforms, satellites, and ground stations and generates snow-related data, such as SWE and snow depth at 1 km resolution. In this study, each of these products has been individually assimilated into Hydrotel, a physically based distributed hydrological model equipped with a snow module, to improve its SWE and spring runoff estimation over the Au Saumon watershed located in Southern Quebec, Canada, for the 2014-2015 water year. The contribution of these products in enhancing spring flow prediction was then evaluated and compared. Our results indicate that both products have the potential to improve runoff prediction in snow-dominated regions. Keywords: Data Assimilation, SWE, Spring Flow, ERA5-land, SNODAS