Spatial variability of snow precipitation and accumulation in COSMO–WRF simulations and radar estimations over complex terrain

Abstract

Abstract. Snow distribution in complex alpine terrain and its evolution in the future climate is important in a variety of applications including hydro-power, avalanche forecasting and fresh water resources. However, the relative importance of processes such as cloud-dynamics and pure particle-flow interactions is still barely known and models are essential to investigate these processes. Here, we present very high resolution Weather Research and Forecasting model (WRF) simulations, which are initialized by 2.2 km resolution Consortium for Small-scale Modeling (COSMO) reanalysis (COSMO–WRF). To assess the ability of COSMO–WRF to represent spatial snow precipitation patterns, they are validated against operational weather radar measurements. Estimated WRF precipitation is generally higher than estimated radar precipitation, most likely due to an overestimation of orographic precipitation in the model. The high precipitation also leads to a higher spatial variability in the model at the scale of 10 km. Overall, an autocorrelation and scale analysis of radar and WRF precipitation patterns show that WRF captures the variability relative to the domain wide variability of precipitation patterns down to the scale of few kilometers, but misses quite substantial variability on the smallest scales of a few 100 meters. However, differences of precipitation between 2830 m above sea level and the ground indicate that near-surface processes are active in the model.

The Importance of Near-Surface Winter Precipitation Processes in Complex Alpine Terrain

“…The simulations are run with initial and boundary conditions from the COSMO-2 analysis by MeteoSwiss. The simulation setup corresponds to the one described in Gerber et al (2018a). All simulations have four one-way nested domains (d01-d04) with horizontal grid spacings of 1350, 450, 150, and 50 m, respectively (Fig.…”

“…The 31 January 2016 and 5 March 2016 dates are chosen because these were two significant precipitation events within the time period, for which photogrammetrically determined airborne snow depth measurements are available (section 2e). Additionally, simulations for these two dates showed better performance compared to simulations for another precipitation event on 4 February 2016 (Gerber et al 2018a). For the variability analysis with respect to radar estimates, the simulations were chosen to cover full calendar days corresponding to the preprocessing of radar measurements.…”

“…Land surface processes are parameterized by the Noah land surface model with multiparameterization options (Noah-MP; Niu et al 2011;Yang et al 2011) and linked to the atmosphere by the MM5 Monin-Obukhov surface layer parameterization (Paulson 1970;Dyer and Hicks 1970;Webb 1970;Zhang and Anthes 1982;Beljaars 1995). Terrain smoothing (four smoothing cycles using the 1-2-1 smoothing option of the WRF preprocessing system; Gerber et al 2018a) is applied to keep all slope angles in domain d04 below 458. This smoothing is necessary to avoid pressure gradient errors (De Wekker 2002).…”

“…The same number of smoothing cycles is applied for all four nested domains to keep topography consistent across domain boundaries. To take missing topography and missing roughness elements into account, the snow surface roughness length is increased to 0.2 m (Gerber et al 2018a).…”

“…Regional simulations are usually limited to a horizontal grid spacing of $1 km as discussed above. A new comparison of Weather Research and Forecasting (WRF) Model simulations driven by Consortium for Small-Scale Modeling (COSMO-2) analysis (COSMO-WRF) to high-resolution radar measurements indicates that wintertime precipitation patterns over complex terrain are not fully captured by model simulations even at a horizontal grid spacing of 450 m, while some additional small-scale variability is resolved at a horizontal grid spacing of 50 m (Gerber et al 2018a). Furthermore, Gerber et al (2018a) found differences in precipitation patterns between a horizontal plane at the elevation of an operational weather radar on a mountain peak and the ground level, indicating the presence of near-surface processes.…”