Evaluation of Snow-Level Height Estimates from Polarimetric WSR-88D Measurements

The use of polarimetric WSR-88D measurements for estimating the snow level (SL) in the vertical column is analyzed. The snow level is defined as an altitude in the atmosphere where precipitation turns from mostly snow to predominantly rain. This level is located within the melting layer, which is typically several hundred meters thick (in vertical). Two WSR-88D SL estimates are evaluated: one is based on the use of the WSR-88D Level III melting layer product, which provides melting layer boundaries for slant beam radar measurements, and the other is based on the location of the minima in copolar correlation coefficient found from the Level-II measurements. The evaluation is performed by intercomparisons of collocated SL estimates from WSR-88D measurements and robust SL data from dedicated vertically pointing snow-level radar data. It is shown that for the close ranges (~ 15 km) the SL estimates obtained from the data contained in the WSR-88D Level III product provide a relatively unbiased values of SL heights with a standard deviation of around 170 m. The copolar coefficient minima heights are biased low by around 100 m lower than SL heights inferred from the vertically pointing SL radars. However, if the average bias is accounted for, the correlation coefficient based SL height estimates are characterized by lower values of the standard deviation compared to the estimates inferred from the Level III data. For longer radar ranges the uncertainties of WSR-88D based SL estimates increase.