Extended Abstract (76K)7B.1
Quantitative Precipitation Forecast for the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPStm)
Sue Chen, NRL, Monterey, CA; and J. E. Nachamkin, J. M. Schmidt, and C. S. Liou
A quantitative precipitation verification (QPF) software has been developed for the Naval Research laboratory's mesoscale model COAMPStm. Evaluation of COAMPStm QPF has been conducted for the winter 2002 season over the CONUS region using the River and Forecasting Center 24 hours accumulated rain gauge analysis (4 km resolution) from National Center for Environmental Prediction. Traditional techniques, such as the equitable threat and bias scores, along with a more recent object oriented verification technique are used to validate COAMPStm QPF skill. The averaged equitable threat and bias scores provide useful guidance on identifying problems of model bias in precipitation but provide little information to the user as to where such biases may exist geographically on the model grid. Additional information in this regard can be drawn from the statistics by plotting the regional frequency distribution for each precipitation category over an extended period of time (in our case three months). Horizontal plots of reginal distribution helps to isolate geographical areas that have the greatest impact on the overall threat and bias scores. Such information can be useful in identifying weaknesses in the model physics, which directly impact the QPF. Some preliminary results showed COAMPS QPF biases are close to one for most of the rain thresholds except for the very heavy precipitation categories (75 mm/day and above). The distribution plots suggest that model over-forecast the precipitation along Pacific Northwest. Another region of low equitable threat scores for the 35 mm/day and above rain thresholds were found to result from consistently under-forecasting the precipitation over the lower Mississippi River Basin. For 50 mm/day and above rain thresholds, less than 20% of the total precipitation was contributed by the COAMPS tm convective scheme suggesting one possible source for the model discrepancy with the onservations.
Session 7B, Statistical Evaluation II
Wednesday, 14 August 2002, 10:30 AM-12:00 PM
Next paper