Abstract
Accurate snow depth datasets are crucial for water resource management, comprehensive climate change evaluations, and the sustainable advancement of the ice-and-snow economy in the context of rapid climate change. To create a high-resolution monthly snow depth dataset tailored for the Northern Hemisphere winter months (NHMSD), this study employed the Delta statistical downscaling method, in conjunction with a spatial feature transfer technique, to refine snow depth data derived from 21 major general circulation models and four shared socioeconomic pathways sourced from the CMIP6 project. The NHMSD stands as the world's pioneering long-term 0.05° snow depth dataset, encompassing the historical era from 1980 to 2014 and extending into future projections from 2015 to 2100. Validation using 2062 ground snow depth observations has confirmed that NHMSD outperforms reanalysis datasets, including ERA5-Land and GLDAS, in terms of root mean square error, bias, and mean absolute error for the periods 1980-2014 and 2015-2023. This dataset facilitates the exploration of potential alterations in snow depth under future scenarios in the Northern Hemisphere.