12A.1 Evaluation of QPE, QPN, and QPF from the 2014 Colorado Front Range Hydrometeorology Demonstration

Thursday, 17 September 2015: 4:30 PM
University AB (Embassy Suites Hotel and Conference Center )
James W. Wilson, NCAR, Boulder, CO; and R. D. Roberts, J. Sun, K. Ikeda, and M. Dixon

From 7 July until 31 August 2015 a real-time demonstration of Quantitative Precipitation Estimates (QPE), Quantitative Precipitation Nowcasts (QPN) and Quantitative Precipitation Forecasts (QPF) where made over eastern Colorado which included the eastern slope of the Rocky Mountains. This paper reports on three statistical evaluations that were conducted. Only data obtained in real-time are utilized and the emphasis is on nowcasting heavy rainfall episodes on the nowcasting time and space scales. The first evaluation utilized dense rain gauge networks to obtain hourly averaged rain amounts over two areas, one for a 100 sq km area over metropolitan Denver and a second 500 sq km area along the Boulder County foothills. These gauge-derived, hourly-averaged rain amounts were then compared with rain amounts obtained from the MRMS, Stage IV, NEXRAD polarimetric and NCAR Polarimetric QPE methods. The second evaluation compared hourly-averaged, rainfall amounts from the MRMS with forecast rain amounts obtained from NWP model precipitation forecasts, radar extrapolation and the NCAR AutoNowcaster. The average rainfall amounts were computed over the two areas mentioned above plus a third 11,000 sq km area for a 60 km radius circle of the Denver KFTG Nexrad. Two of the NWP models did not include radar data assimilation (GFS and 3DVAR) and 4 NWP models did include radar data assimilation (3DVAR, 4DVAR, RTFDDA and ADAS). The third evaluation was a grid point (1 km) to grid point comparison of the MRMS 1, 2 and 3 hr rainfall amounts with those from the NWP model forecasts for a variety of forecast lead times, precipitation thresholds and grid relaxations. The ongoing analysis is showing interesting differences between 1) the various QPE methods; and 2) the effect of radar data assimilation and forecast lead time on the accuracy of forecast rainfall. Full findings will be reported on at the conference.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner