Verification of Hydrometeor Classification Algorithms for the U.K. C-band Radar Network

Knowledge of precipitation type in near-real-time has many applications in both operational and research meteorology. Assimilation into NWP models and precipitation microphysics research are examples of uses of such information. Hydrometeor classification using weather radar is particularly desirable, due to radar’s high temporal and spatial resolution. Over the past 3 years the United Kingdom Met Office network of C-band Doppler weather radars has been upgraded to dual-polarisation which allows characteristics of hydrometeor phase, shape, and orientation to be determined. This information can be used as input in classification algorithms (i.e. fuzzy logic) to determine the most likely hydrometeor type within the observed volume. Despite such algorithms being used operationally for over a decade worldwide (notably the USA), there is limited rigorous verification in the literature. Verification of hydrometeor classification skill is a particular challenge because both the classifier and the verification data are non-discreet. In addition, the majority of precipitation observations are made on the ground, which creates additional sources of uncertainty as the particles fall from radar-height to the surface.

This PhD project aims to build a rigorous framework by which to verify radar hydrometeor classifiers, enabling multiple algorithms to be compared with a wide array of data sources that have various levels of certainty and representation. Using these novel methods, the improvement in hydrometeor classification using dual-polarisation radar data compared to what was previously possible with single polarisation radar will be demonstrated. To aid this process a network of 13 laser disdrometers across the UK (DiVeN) has been recently established for this project. These instruments can measure the fall speed and diameter of particles that fall through the instrument’s laser beam and this can be used to empirically deduce the precipitation type. Here we present initial hydrometeteor class verification results using the disdrometer data and other sources of observations.