Methods for Identifying Systematic Differential Reflectivity (Zdr) Biases on the Operational WSR-88D Network

A dual-polarization upgrade to all 143 operational Weather Surveillance Radar-1988D (WSR-88D) radars in the contiguous United States (CONUS) was completed in May 2013. This upgrade makes several important radar variables available to end users and meteorological algorithms. Differential reflectivity (Zdr), one of several new dual-polarization variables, gives hydrometeor shape information and aids in distinguishing hydrometeor type and making quantitative precipitation estimates (QPE). The hydrometeor classification algorithm (HCA), melting layer detection algorithm (MLDA), and QPE algorithm employed by the WSR-88D radar product generator requires unbiased Zdr data to achieve the best performance. We are focused on improving Zdr calibration procedures because an absolute Zdr bias of 0.1 to 0.2 dB can produce a QPE error of 10-30% in areas of rain not contaminated by hail or mixed phase precipitation.

The tri-agency Radar Operations Center (ROC) actively monitors fleet wide WSR-88D Zdr biases through a variety of means. First, the WSR-88D system utilizes the manufacturer's recommended engineering method for calibrating the radar system hardware. A system Zdr bias estimate is generated through this process and is applied to all final Zdr values. Presently, this is the only dynamic engineering method for which a system bias correction is applied to Zdr values. The ROC is also pursuing additional engineering methods for monitoring or adjusting system Zdr bias, such as cross-polar power calibration. Second, the ROC monitors system Zdr bias using an automated weather method originally developed by at the National Severe Storms Laboratory (NSSL). To use the NSSL method, the ROC applies innovative filtering to isolate stratiform precipitation, reducing the likelihood that system Zdr bias values are heavily influenced by variability in the drop-size distribution. Third, the ROC monitors Zdr bias values derived from daily sunspikes (archived in Level II data) to monitor receiver and antenna bias contributions to total system Zdr bias. Finally, the ROC makes use of the ROC's Hotline maintenance and system status logs to correlate Zdr bias changes tied to maintenance actions.

When data from all these methods are overlaid graphically, systematic Zdr biases become obvious. The ROC successfully used these methods to identify maintenance and procedural issues at several sites and followed through with corrective action. Preliminary results for January to March 2013 (194 stratiform events from 64 unique WSR-88D sites) indicate that approximately 70% of events had an absolute system Zdr bias ≤ 0.2 dB. The remaining 30% of events had an absolute system Zdr bias of > 0.2 dB. This paper summarizes our evolving Zdr calibration monitoring methods and lessons learned regarding systematic Zdr biases on the WSR-88D fleet.