Complementarity between snow remote sensing, gauges, radar observations and Numerical Weather Prediction models to better constrain solid precipitation accumulation in spatially distributed snow cover modelling

Accurate spatially distributed simulations of snow cover in mountainous regions is highly dependent on the possibility to well constrain the accumulation of solid precipitation. A number of observations and model data can provide direct or indirect assessment of their amount with varying spatial resolutions, spatial coverage and uncertainties. However, the complementarity between the different sources of informations is poorly documented and methodologies to appropriately combine all data are missing.

In this work, we present a new modelling framework taking benefit from (1) radar observations of precipitation, (2), local precipitation gauges, (3) precipitation climatology of a Numerical Weather Prediction model and (4) satellite remote sensing of snow depth. We show over a 900 km² simulation domain in Central French Alps that all data sources help to better constrain precipitation and to obtain more realistic snow depth spatial patterns. Radar observations provide the best temporal chronology of precipitation but the NWP model help to capture better altitudinal and horizontal climatological gradients and to fix spatial artefacts in radar measurements due to ground clutter. The assimilation of satellite snow depth maps is found as highly beneficial to capture spatial patterns of accumulated solid precipitation not well captured by radars and NWP. The added value of snow depth maps is maintained several months after the assimilation date, but they can not solve the errors specific to individual precipitation events. As a result, the most realistic spatial patterns of simulated snow depths are obtained when all sources of data are combined, with appropriate ensemble algorithms and uncertainty quantification.

Finally, we outline short term perspectives to integrate real-time snow observations from optical satellites in the previously described framework. This is an important step in the development of the EDELWEISS high-resolution (250 m) snow modelling system, which is expected to cover all French mountains by 2026.