866 A One Year Assessment of Four Radar-Based QPEs across the Continental US

Wednesday, 9 January 2019
Hall 4 (Phoenix Convention Center - West and North Buildings)
Stephen B. Cocks, CIMMS/Univ. of Oklahoma and NOAA/NSSL, Norman, OK; and L. Tang, S. M. Martinaitis, J. Zhang, A. V. Ryzhkov, P. Zhang, and K. W. Howard

A year-long assessment of radar only quantitative precipitation estimates (QPE’s) across the contiguous US (CONUS) was recently completed. Data was collected from 111 WSR–88D radars on 219 calendar days between April 2017 and April 2018, a total of 569 precipitation events. The collected data was divided into regions east and west of the Rocky Mountains and evaluated for the warm (April - September 2017) and cool (October 2017 - April 2018) seasons. The radar based QPEs evaluated were the operational Dual Pol, legacy precipitation processing system (PPS), multi-radar multi-sensor (MRMS) radar only (Q3RAD) and the developmental MRMS Dual Pol (Q3DP). Dual Pol primarily uses R(Z,Zdr), R(Kdp) and R(Z) dependent upon the radar echo class given by the hybrid hydro-meteorological classification algorithm (HHC) with some bright band (BB) mitigation applied within the melting layer. PPS uses a single R(Z) relation whereas Q3RAD uses multiple R(Z) relations dependent upon the radar echo classification of its surface precipitation type algorithm which utilizes BB corrected reflectivity and model data. Q3DP utilized rain rates derived via specific attenuation below the melting layer, specific differential phase for where hail was likely and BB corrected reflectivity within and above the melting layer.

The results showed that for the period evaluated, the operational Dual Pol QPE exhibited a significant underestimate bias in both the Western and Eastern US. Further, while the operational Dual Pol QPE performance during the Western US warm season was better than the PPS and Q3RAD QPEs, its overall performance during the Eastern/Western cool and Eastern warm seasons was similar to PPS. Using Radar Operations Center methods for quantifying the amount of differential reflectivity (Zdr) bias, the operational Dual Pol QPE performance assessment was re-accomplished using data collected only from radars where Zdr was considered to be well calibrated (|Zdr bias| ≤ 0.2) but the results did not appreciably change. Overall, the mosaicked MRMS Q3RAD and the developmental Q3DP products generally outperformed the single radar QPEs. MRMS Q3DP exhibited less variability and significantly lower error than Q3RAD, particularly during moderate to extreme rainfall events.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner