| Description |
We explore the ability of in-situ snowflake microphysical observations to constrain estimates of surface snowfall accumulations derived from co-incident, groundbased radar observations. Specifically, in winter 2016-17, we deployed a Micro Rain Radar (MRR), Precipitation Imaging Package (PIP), and Multi-Angle Snow Camera (MASC) to the Haukeliseter Test Site run by the Norwegian Meteorological Institute. This state-of-the-art snowfall measurement station lies near an elevation of 1000 m on a plateau in the coastal mountains of Norway and typically receives up to 2-3 m of snow per year. The site also houses a double fenced snow gauge and a comprehensive set of basic meteorological observations. In terms of the retrieval work, observations of snowflake particle size distribution (PSD), fall speed, and habit from the PIP and MASC were used as input to a snowfall retrieval algorithm based upon MRR measurements. Retrieved snowfall accumulations were then compared with double fenced snow gauge observations to evaluate retrieval performance as a function of storm event type and meteorological conditions for the Haukeliseter site. These analyses found differences of less than 10% between snow gauge and MRR-retrieved estimates of total seasonal accumulation when using either PIP or MASC observations for low wind, moderate reflectivity ‘upslope' snow events. Agreement was less robust for the high wind, high reflectivity ‘pulsed' snow events often observed at the site. Differences in total seasonal accumulations for these events were 49% and 59% when using MASC and PIP in-situ observations, respectively. During these ‘pulsed' events, blowing snow produced by high winds and measured by the ground-based instrumentation was likely not representative of the snowfall aloft as observed by the MRR, thus biasing retrieval results. However, the use of alternate assumptions of MRR Doppler velocity and temperature-based estimates of snowflake PSD for these ‘pulsed' storms produced a better seasonal agreement with a difference of 16% between snow gauge and MRR-retrieved estimates. Such results suggest that the MRR-PIP-MASC snowfall retrieval scheme has potential for estimating snowfall rate but that the combination of ideal measurements for the scheme will depend upon specific meteorological conditions. |
