Recorded presentation8.3
Validation of the NWS Stage IV Multi-Sensor Precipitation Estimates (MPE) in Mid-Atlantic Region
Boone F. Larson, University of Louisiana at Lafayette, Lafayette, LA; and A. Tokay, E. Habib, and B. R. Nelson
Multi-Sensor Precipitation Estimates (MPE) is a regional multi-sensor hourly surface rainfall product that was developed for input into hydrologic forecast models and decision-making systems for river forecasting, flood and flash flood warning, and other hydrologic monitoring purposes. MPE result from merging operational radar, automated gauge, and Geo-stationary satellite rainfall estimates following quality control and bias adjustment. MPE are superimposed over a nominal grid size of 4 square kilometers termed Hydrologic Rainfall Analysis Project (HRAP). Stage IV is a final stage term used to describe nationwide mosaicking of manually-edited, regional MPE products produced by each of River Forecast Center (RFC) on an hourly basis. Stage IV is a readily available operational product which makes it attractive to validate various satellite rainfall estimates including NASA's Tropical Rainfall Measuring Mission (TRMM) multi-sensor precipitation analysis. Prior to the use of Stage IV as a validation product, the accuracy of Stage IV products should be validated with independent observations. In that regard, this study compares Stage IV rainfall with a rain gauge network that was located in Eastern Virginia, Delmarva, and northeast North Carolina. The rain gauge network consisted of 28 sites where each site had dual or triple gauges. Interestingly, three-fourths of the sites were at the coast. In fact, one-fourth of the sites were located at Wallops Island, Virginia where the maximum site spacing was approximately 6 km. As a result, two MPE pixels each had three gauge sites at Wallops Island, while the rest of the MPE pixels had only a single gauge site. The gauges were operated by the TRMM satellite validation office during 2004-2007 and were not in included in Stage IV. Each site was in operation at least a year and quality control of the gauges has been completed as part of this and previous study. A rainfall statistical package including correlation coefficients and biases were run between gauge and MPE rainfall for one-, three-, and six-hour, daily and monthly time scales. The statistical package was also applied between gauges and neighboring eight MPE pixels, gauge and average of nine MPE pixels, and between the nine MPE pixels themselves. The preliminary results indicate that agreement between MPE and gauge rainfall is reasonable at a given site on an hourly basis, and agreements between the two increase with increasing integration time scale. Several thresholds were also applied to the gauge rain totals to emphasize higher rain totals which are significant in hydrological applications. Based on our knowledge, the evaluation of Stage IV or previous versions (Stage I, II, III) has been done at other locations, but not in a coastal area and where rainfall can be dominated by tropical cyclones, frontal systems, and isolated thunderstorms. Recorded presentation
Session 8, Validation of Hydrometeorological Observations, Part II
Wednesday, 23 January 2008, 10:30 AM-12:00 PM, 223
Previous paper Next paper