Application of the CLEAN-AP Clutter Filter using WET for Improved Quantitative Precipitation Estimation

A major source of errors in quantitative precipitation estimation (QPE) comes from the fact that radar measurements are made aloft, whereas rainfall estimates at ground level are the quantity of interest. While the Met Office makes corrections for the vertical profile of reflectivity (VPR) in order to better estimate surface precipitation rates, it is known that the errors associated with this correction increase with altitude.

The application of ground clutter filters allow the use of more low elevation data than would be possible otherwise. However, it is known that in some meteorological conditions, ground clutter filtering can bias the reflectivity measurements. This can particularly be an issue when dealing with the low maximum unambiguous velocities associated with C-band rain-rate optimised pulse repetition frequencies (PRF).

As part of the collaboration between the Met Office Radar Systems team and the National Severe Storms Laboratory (NSSL) Advanced Radar Techniques (ART) team, the Clutter Environment Analysis using Adaptive Processing (CLEAN-AP) algorithm (Torres and Warde 2014) has been implemented as the ground clutter filter on the Met Office Cyclops-D signal-processing system. Work has been carried out to optimise the filtering at C-band. In addition, the Weather Environment Thresholding (WET) algorithm (Warde and Torres 2015), used to select when it is appropriate to use unfiltered data and when it is preferable to use the CLEAN-AP filtered data, has also been implemented in the Cyclops-D signal-processing system.

In this work, we present means of generating I&Q data with known characteristics for the evaluation of the CLEAN-AP and WET algorithms. While simulation frameworks do exist to produce realistic I&Q data for clutter and weather-like signals, these do not capture the spatial structure that is essential to selection algorithms such as WET. Using I&Q data from known weather and ground clutter cases, we combine these to generate raw data that has realistic spatial characteristics, but with known properties.

We go on to demonstrate the impact of effective clutter mitigation and selection on downstream products and QPE. This is done by looking at the impact on the spatial smoothness of accumulated reflectivity fields, visual evaluations of Doppler and dual polarisation parameters, the smoothness of rainfall accumulations, and on comparisons of radar precipitation estimates against gauges.