Abstract Snowfall data from ERA5, GPCP V3.2, IMERG V07, and IMERG V06 were assessed over Arctic sea ice by reconstructing snow depth from these products and comparing the results to reference snow-depth values derived from combined ICESat-2 and CryoSat-2 observations across three snow accumulation seasons (2018–2021). Spatiotemporal analyses suggest that ERA5, GPCP V3.2, and IMERG V06 exhibit pattern agreement with the reference snow depth over the central Arctic, with correlations up to 0.84. IMERG V06 significantly underestimates snow depth and shows relatively low spatiotemporal variability. We found that IMERG V07 has lower skill in capturing the spatial distribution of snow depth compared to IMERG V06 at the expense of reducing overall bias and increasing spatial and inter-month variability. This work marks an initial yet significant step toward a more integrated assessment of snowfall and snow depth estimates over sea ice, where in situ observations are largely absent.
Song et al. (Thu,) studied this question.