Validating WRF model precipitation phase forecasts using vertical Doppler profilers and disdrometer observations

Abstract

In this work we aim to validate data from two model runs of the Servei Meteorològic de Catalunya’s WRF model post process using the data of three co-located instruments. Theseinstruments are two Micro Rain Radars (MRR2 and MRR-Pro) and a disdrometer (OTT Parsivel 2). Two Fuzzy Verification approaches to resample the data and validate the forecasts were used: one considers the predominant precipitation type during a one-hour interval while the other includes all types which occurred during this period. Contingency Tables were made and Probability of Detection (POD), False Alarm Ratio (FAR) and Gilbert Skill Score (GSS) were used to analyse the model behaviour against lead time and UTC. Results show that the model’s performance depends strongly on lead time and on the method used to resample the data. Tendencies were observed comparing both model runs which are discussed considering the limitations of the study