Mountain snow remains a tremendous challenge to measure or model. Historically, scientists studying fallen vs. falling snow have operated within separate communities, to our detriment. Uncertainty in snowfall is currently the largest error in modeling snow on the ground (Raleigh et al. 2016), while unknown snow on the ground poses significant challenges to the remote sensing of snowfall. GPM-based IMERG estimates currently produce only about half of the observed snow in the Olympics (Cao et al. 2018), and GPM retrievals underestimate snowfall in the Alps (Speirs et al. 2017). Therefore improving representations of total annual snowfall magnitudes are the most critical first step to improving our understanding of snow's lifecycle.

We Hypothesize:

  1. High-resolution mesoscale models can be used to constrain cloud microphysics as a function of both the large-scale atmospheric environment and the land surface topography to meaningfully constrain satellite retrieval algorithms. Knowledge from well-observed sites (e.g., those with radars and lidar-based snow depths) can be expanded to global scales where large-scale environmental variables, e.g., from operational weather prediction models and land surface topography, are available.
  2. Patterns of snowfall in complex terrain are consistent enough that including prior information regarding these patterns (from historic snow depth and disappearance date data) can be used to extend GPM snowfall estimates from regions of high confidence outside of complex terrain, to regions in complex terrain where GPM’s radiometer information has demonstrated only limited skill to date.
  3. Snow depth retrievals along a narrow strip, as can be achieved by space-borne lidar (e.g. ICESat-2 and GEDI), can be assimilated in conjunction with GPM-based snowfall and historic snow depth patterns to improve range-wide estimates of snow. We will initially test our hypotheses in well-observed intensive study areas, including the focus areas of OLYMPEX (Olympic Mountains in Western Washington), ASO (the Sierra Nevada in California), and SnowEx 2017 and 2020 (Colorado and the Western U.S.). We will then use global models and reanalysis datasets, combined with MODIS- and Landsat-based historic snow reconstruction patterns, to expand the methodology globally. We propose a stepwise approach to bridge the gap - at each step, correcting what can be corrected, while framing the uncorrectable as an uncertainty that can only be addressed by better measurements and/or models.