The algorithm used in the SPT product uses 1) radar estimated precipitation rate and 2) the local height of the 0ºc wet-bulb isotherm from an NWP model. Therefore, the SPT product attempts to resolve precipitation type changes over terrain and also capture the role of evaporative cooling to more accurately detect the rain-mixed-snow boundary. Hail is also detected with a reflectivity-height criterion.
In this study, 5 years of SPT data are compared against several surface in-situ measurements: 1) 106 UK Met Office automatic present weather stations, which rely on a visibility-temperature algorithm; 2) 33 UK Met Office stations manned by a trained meteorological observer; 3) a network of 13 optical disdrometers installed around the UK (the Disdrometer Verification Network, or DiVeN); 4) public crowdsourced reports of hail which are quality-controlled by the European Severe Weather Database (ESWD).
Both the SPT product and the in-situ surface data are categorical, and both have deficiencies which make the verification complex. These issues are addressed using a verification technique which varies the spatial and temporal tolerances of the SPT product, amongst others. The results not only evaluate the SPT product; they also highlight the strengths and weaknesses of different surface observations for precipitation type measurements.