Downscaling Low-Resolution Snow Water Equivalent from Passive Microwave Data for Catchment Observations - Comparison of Two Approaches

Over 20% of the world's population depends on snow as a freshwater resource, whether for drinking water, agriculture, or hydropower. However, climate change is putting snow and its role as a freshwater reservoir under increasing pressure due to changing precipitation patterns and altered seasonality. This makes snow less reliable for various communities, economies, and ecosystems, including the corresponding flora and fauna. To continue utilizing snow as a water source, these communities rely on accurate and reliable data on snow water equivalent (SWE).

Despite improvements in SWE data over recent decades, global remote sensing datasets—primarily based on passive microwave (PMW) measurements—still face significant challenges. One major issue is their low spatial resolution (25x25 km), which limits their applicability in complex topographical regions and small-scale studies that are crucial for local communities and economies. On the other hand, data on snow cover area (SCA), typically derived from optical or SAR measurements, offer increased spatial resolutions but lack information about the snow's water content. In addition, radar can only detect wet snow, and optical instruments only work in daylight and without clouds.

To bridge the information gaps between SWE and SCA data, we utilized our MODIS/VIIRS-based Global Snow Pack (GSP), an SCA dataset with a resolution of 375 m. Within the GSP processor, the snow cover information is merged from three sensors (MODIS Terra & Aqua, Suomi-NPP VIIRS), and via four consecutive interpolation steps, cloudless daily snow coverage is obtained globally. With this daily snow cover, one can determine the number of days during which each pixel is continuously covered with snow. This information is used to downscale the low-resolution PMW data in a weighted manner for the years 2000 to 2024. However, an alternative approach was also investigated. Using shapefiles from different catchments, analysis grids were created that take into account the percentage share of the catchment area under consideration in the case of overlapping SWE pixels - especially for smaller catchments. Both methods are compared and validated.

For validation, we utilized local station-based SWE and river runoff data. We compared basins of different sizes on three continents (North America, Europe, Asia) with varying proportions of mountainous terrain, as we expect not only the resolution but also the saturation effects of the PMW sensors to influence accuracy in these regions. The improved resolution of the SWE data can enhance local snow and streamflow analyses based on remote sensing data. This can be assessed by comparing streamflow predictions derived from station data, original PMW data, and our new dataset. Detailed preliminary results will be presented at the conference.