The impact of temporal aggregation of solid precipitation measurements on the correction for wind induced undercatch.

Solid precipitation measurements are strongly affected by wind-induced errors, due to the aerodynamic response of catching type gauges. The snowflakes deviate from their undisturbed trajectories due to the alteration of the airflow field around the body of the gauge and the corresponding developed turbulence. The resulting effect consists in a certain degree of undercatch, which is a function of the undisturbed wind velocity. Windshields have been developed to provide reference solid precipitation measurements in field test sites, e.g. the Double Fence Intercomparison Reference (DFIR), or employed in operational measurements with the aim of reducing the effect of the wind, such as the Single Alter shield (SA). Various authors have studied wind-induced errors from different point of view (theoretical, numerical and experimental). The available data from CFD and field tests studies (Colli, 2015; Wolff,2 2015) lead to determine the Collection Efficiency (CE) of the gauges, which is a relation between the solid precipitation undercatch and the wind speed, and consequently the transfer function (TF) used to correct the wind-induced errors. Since solid precipitation records, and the associated wind speed data, are commonly stored with a quite coarse resolution in time (60 or 30 minutes), we investigated the impact of the aggregation scale on the accuracy of snowfall data when corrected by using the transfer functions. From the WMO SPICE (Solid Precipitation Intercomparison Experiment) field campaign, we selected a number of snowfall events from the Marshall Field test site (Colorado, USA) during the winter seasons from 2013 to 2015. Three Geonor weighing gauges with different configurations (unshielded, Single Alter shielded and DFIR as the reference) recorded the snowfall rates. Both precipitation and wind speed data are quality controlled and provided with the time resolution of 1 minute. The CE and the respective Transfer Function have been derived from the selected number of snowfall events. From the reference (DFIR) snowfall data and the wind speed values at the time resolution of 1 min, we derived synthetic sequences of wind affected snowfall rates using the derived CE curves at the coarser time resolution. The synthetic observations have been obtained both for the Unshielded and SA gauges. Using the TF, the synthetic dataset has been corrected using different aggregation intervals (5, 15 and 30 minutes). It is observed that, by increasing the aggregation interval of both snow and wind data, the efficacy of the correction decreases, indicating that high resolution snowfall and wind measurements are essential to ensure high quality snowfall measurements.