Improving the Detection Probability of Low Clutter-to-Signal Ratio Ground Clutter Contamination in the WSR-88D

The Clutter Mitigation Decision (CMD) algorithm is a fuzzy-logic-based system that automatically detects ground clutter contamination in the WSR-88D radars. Identified regions then have the clutter filter applied to them to remove the clutter echo. The most difficult aspect of clutter identification is in mixed clutter and hydrometeor echoes, particularly as the clutter-to-signal-ratio (CSR) becomes more negative, i.e. the underlying clutter signals are weaker than the hydrometeor signals. The CMD algorithm has been shown to do well at removing clutter bias in the base moments of reflectivity (Z), radial velocity (Vr) and spectrum width (W). However the dual-polarimetric variables are biased by clutter at much lower CSR values than are Z, Vr and W, as has been shown in several studies. While CMD is able to detect the vast majority of clutter bias of Z, Vr and W, there is considerable undetected contamination in low CSR regions in the variables correlation coefficient (RHOHV), differential reflectivity (Zdr) and differential phase shift (PHIDP) due to their high sensitivity to underlying clutter. It was found that adjusting the existing CMD algorithm in order to increase the detection of clutter in regions with these low CSR values increased the false alarms in pure weather to an unacceptable level. Therefore a new approach to identifying clutter contamination at low CSR was needed.

In order to solve the low-CSR clutter detection problem, we propose comparing RHOHV values before and after the clutter filter has been applied (RHOHV_unfilt and RHOHV_filt, respectively). This RHOHV-test is based on the observation that in regions of mixed clutter and hydrometeor echo the removal of the clutter echo results in the measured RHOHV moving substantially closer to 1.0. Therefore the degree of positive change in RHOHV_filt from RHOHV_unfilt can be used as an indicator of low CSR clutter. A RHOHV improvement factor is therefore computed that indicates how much closer to 1.0 RHOHV_filt is than RHOHV_unfilt. If the RHOHV improvement factor exceeds an a priori threshold, currently a factor of 4.0, the dual-polarimetric data are flagged for filtering. The RHOHV-test is an additional step to CMD.

This type of test is only valid using RHOHV, however the clutter detection results of RHOHV-test are applicable to RHOHV, Zdr and PHIDP. It was found using that RHOHV-test to identify low CSR clutter in addition to CMD dramatically decreased clutter contamination in RHOHV, Zdr and PHIDP. However using the RHOHV-test output to filter Z, Vr and W, increased the clutter filter bias. Thus the additional RHOHV-test is recommended for use only on the dual-polarimetric variables.