Using snow depth data to minimize the error in winter precipitation measurements

In mountain regions like the Alps a significant fraction of the annual precipitation falls as snow. The timing of water resource formation and meltwater release from this snow cover have major implications on the hydrology of mountain catchments with various ecological, social and economic aspects. Operational services, such as traffic maintenance, real-time flood-warning systems of hydrological services and avalanche warning, but also hydropower companies and ski resorts require accurate predictions of precipitation, snow depth and the corresponding snow water equivalent of snowfall. In general, there is a growing need for precipitation data in a high temporal and spatial resolution. However, producing accurate precipitation maps in complex terrain using only remote sensing techniques and rain gauge data is a difficult task. Precipitation data suffer from error margins of up to 50% because of undercatch of solid precipitation in standard rain gauges. Taking snow depths instead of rain gauge measurements seems advantageous, but the density of new snow has to be known, measured or calculated to estimate water equivalent of snowfall from measured snow depths, whereas the spatial and temporal variability of new snow density is not fully understood. The project pluSnow aims to combine high-accuracy snow depth measurements and precipitation data in a high temporal resolution on a sub-hourly scale to investigate in detail the relation between snow depths and gauged precipitation and to minimize the error of gauge undercatch. Key investigations concern i) the difference between the water equivalent of snowfall calculated from the height of new snow and gauged winter precipitation in a high temporal resolution, ii) the relationship between new snow densities and meteorological parameters from station data and gridded weather analysis, and iii) improvement of winter precipitation measurements by assimilation of measured snow depths. The analysis is performed on the basis of snow depth data from 55 automatic weather stations (TAWES), operated by the Austrian Central Institute for Meteorology and Geodynamics (ZAMG) and equipped with laser sensors to measure snow depth with high accuracy. New snow densities are investigated on the basis of high-quality snow density data and meteorological observations from 4 automatic weather stations, which are carefully distributed to capture different weather conditions in the Alps. The pluSnow project will contribute to existing research efforts around the globe which focus on improving the precision of solid precipitation measurements. Results from the proposed project help to fill gaps in the knowledge about Alpine climatology with a special focus on winter precipitation. The investigation will help to minimize errors of conventional pluviometer measurements by assimilation of measured snow depths in precipitation analysis. In this way, the results will improve gridded analysis products as well as nowcast and forecast of snow depths and precipitation.

The project pluSnow aims to combine high-accuracy snow depth measurements and precipitation data in a high temporal resolution to investigate in detail the relation between snow depths and gauged precipitation. As a first result, a high correlation between precipitation and the height of new snow could be detected using hourly and sub-hourly data of the optical snow depth sensor SHM30 (G. Lufft Mess- und Regeltechnik GmbH). This sensor is mounted at more than 80 automatic weather stations operated by Zentralanstalt für Meteorologie und Geodynamik (ZAMG) and provides snow depths with an accuracy on the mm-scale also during snowfall events. The density of new snow has to be known to convert the height of new snow into liquid precipitation. An average density of the new snow for hourly snowfall data of 68 9 kgm-3 was calculated from the combined analysis of snow depth data and snow water equivalent measurements from four mountain stations. This mean new snow density is considerably lower than the approximation of 100 kgm-3 often used for daily snowfall amounts. The hourly variations in new snow density could not be explained in a satisfactory manner using meteorological data measured at the station location so far. A comprehensive comparison between the water equivalent of new snow and precipitation corrections as suggested by the WMO (SPICE experiment) over a large set of automatic weather stations shows that the difference between snow depth changes and precipitation data is very site specific. Thus, a global calculation of the correction factors of the winter precipitation by means of optical snow height data and new snow density algorithms is not operationally applicable to all stations, but appears to be useful to correct for undercatch of gauged winter precipitation at a number of automatic weather stations. In general, improvements of snow cover modelling are the basis for applications in operational services, such as traffic maintenance, real-time flood-warning systems of hydrological services and avalanche warning, but also hydropower companies and ski resorts, which all aim in increasing public safety and sustainable use of resources. The project shows that the measurement accuracy of the optical sensor is sufficient to analyse sub-daily snow depth changes for the comparison with precipitation data, and that mean new snow densities can be estimated from snow-hydrological measurements of existing SWE and snow depths. The pluSnow project attributed to the detailed performance analysis and further development of a snow model applied by ZAMG and partners for different applications in Austria.