Exploring the sensitivity to precipitation, blowing snow, and horizontal resolution of the spatial distribution of simulated snow cover

Abstract

Abstract. Accurate snow cover modeling is a high stake for mountain regions. Alpine snow evolution and spatial variability result from a multitude of complex processes including interactions between wind and snow. The SnowPappus blowing snow model was designed to add blowing snow modeling capabilities to the SURFEX/Crocus simulation system for applications across large spatial and temporal extents. This paper presents the very first spatialized evaluation of this simulation system over a 902 km2 domain in the French Alps. Here we compare snow cover simulations to the spatial distribution of snow height obtained from Pleiades satellites stereo-imagery and to Snow Melt-Out Dates from Sentinel-2 time series over three snow seasons. We analyzed the sensitivity of the simulations to three different precipitation datasets and two horizontal resolutions. The evaluations are presented as a function of elevation and landform types. The results show that the SnowPappus model forced with high-resolution wind fields enhances the snow cover spatial variability at high elevations allowing a better agreement with observations above 2500 m and near peaks and ridges. Model improvements are not obvious at low to medium altitudes where precipitation errors are the prevailing uncertainty. Our study illustrates the necessity to consider error contributions from blowing snow, precipitation forcings, and unresolved subgrid variability for robust evaluations of spatialized snow simulations. Despite the significant effect of the unresolved spatial scales of snow transport, 250 m horizontal resolution snow simulations using SnowPappus are found to be a promising avenue for large-scale modeling of alpine snowpacks.

Radar based high resolution ensemble precipitation analyses over the French Alps

“…This will allow the resulting simulated snow properties to be compared with corresponding satellite observations thanks to the memory effect of the snowpack, which ensures a strong relationship between the state of the snowpack and past snowfall (Haddjeri et al, 2023). The main challenge of this indirect evaluation of precipitation analyses will be the introduction of additional uncertainties such as the height of the rain-snow limit (Vionnet et al, 2022) and the errors of the snowpack model (Essery et al, 2013;Lafaysse et al, 2017).…”

“…-The WMA filter described in section 3.1.4 tends to smooth precipitation fields. This is a necessary compromise to remove unrealistic spatial patterns that would be very detrimental to snow modelling (with unrealistic snow amounts near ridges, as illustrated by Haddjeri et al, 2023, with the ANTILOPE raw product). However, it can potentially erase realistic -The climatological approach of this method makes it highly dependent on ANTILOPE evolutions (changes in the algorithm, in the available radar or in the in-situ measurements used), so that regular recalibration would probably be required.…”

“…However, they suffer from biases and positioning errors in individual events and in seasonal accumulations. These errors are problematic for snow cover modelling (Vionnet et al, 2016(Vionnet et al, , 2019Haddjeri et al, 2023). A combination of surface observations and NWP output is used in some operational snow modelling systems (SAFRAN, Durand et al, 1993;Lespinas et al, 2015) to provide precipitation estimates at scales of a few hundred square kilometres.…”

“…Methods have been developed to correct radar-based precipitation fields (Vogl et al, 2012) and to assess the associated uncertainty (Kirstetter et al, 2010;Villarini et al, 2014;Kirstetter et al, 2015). However the low quality of precipitation estimates based on radar measurements in complex terrain, even when combined with in-situ observations (Sideris et al, 2014;Sivasubramaniam et al, 2019;Champeaux et al, 2009), currently prevents their direct use to successfully force a snowpack model (Haddjeri et al, 2023). More sophisticated products combining NWP outputs, surface observations and precipitation estimates from radar measurements (CaPA, Fortin et al, 2015Khedhaouiria et al, 2022) suffer from significant biases in winter (Lespinas et al, 2015).…”

“…The spatial structures of the analysed precipitation fields could not be evaluated with the reference data set used in this study. An indirect evaluation of simulated snow depths based on the precipitation analyses presented here is planned in a future study, following the workflow already proposed byHaddjeri et al (2023). It will compare simulated snow depths with snow depth maps derived from high resolution Pleiades satellite images(Deschamps-Berger et al, 2020) over the Grandes Rousses domain.…”