Session 6R.4 Priliminary Results For The 0-1 Hour Multisensor Precipitation Nowcaster

Wednesday, 26 October 2005: 11:15 AM
Alvarado ABCD (Hotel Albuquerque at Old Town)
Shucai Guan, IMSG at NOAA/NWS/NCEP/NCO, college park, MD; and F. Ding, R. A. Fulton, and D. H. Kitzmiller

Presentation PDF (899.0 kB)

The capability of WSR-88D to estimate rainfall in real time makes it an attractive tool for use in flash flood hydrology. Using the radar data as a primary input, the Multisensor Precipitation Nowcaster (MPN) algorithm can produce very-short-term (0-1 hour) deterministic regional, gridded precipitation forecasts to provide Weather Forecast Offices with additional forecast guidance and lead-time for issuance of flash flood warnings. The NOAA National Weather Service needs to provide automated guidance to forecasters on very short term future rainfall, especially intense rainfall out to one to three hours in the future, to improve the detection, prediction, and warnings for flash floods. This short term 0-3 hour forecast period is currently not well predicted by atmospheric forecast models, and thus forecasts at these short lead times depend on extrapolation of current reflectivity observations into the future. The MPN system ingests data from multiple radar units, improving on the representation of the radar reflectivity field by minimizing radar range degradation effects.

The MPN algorithm uses the WSR-88D Digital Hybrid Scan Reflectivity product as its primary input, estimates radar echo motion vectors using a pattern matching technique, and considers explicitly storm development and decay to forecast short-term precipitation. Forecasts from the MPN have been analyzed for 8 flood cases in MD-DC-VA region using radar data from Sterling, Virginia WSR 88D unit. Several tunable parameters of the MPN algorithm have been tested for these cases: temporal and spatial smoothing, growth/decay functions, and global vs. local extrapolation velocities. Forecasted rain rates from MPN have been compared with observed rain rates from radar during the valid period. The results show that root mean squared error of the rainrate forecasts, valid at 60 minutes, produced by MPN is 10 – 25% smaller than the error associated with the common method of pure extrapolation. The details of MPN and further analysis of its short-term precipitation forecasts will be presented at the conference.

- Indicates paper has been withdrawn from meeting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