Evaluation of the GPCP Rainfall Product via the Surface Reference Data Center

The Surface Reference Data Center (SRDC), a subprogram of the Global Precipitation and Climatology Project (GPCP), collects surface rainfall data to compare with satellite rainfall estimates produced by the GPCP. The satellite estimates are produced on a 2.5ºX2.5º monthly and an experimental 1.0ºX1.0º daily scale. The surface data is collected so that both time and space resolutions can be examined. One of the stated goals of the SRDC is to produce products with known error; thus, an examination of the error characteristics of the data has been undertaken. This includes the development of the standard error of the mean and the signal to noise ratio. This allows for uncertainty bounds to be placed on the surface estimates, therefore allowing SRDC to choose which areas will provide the most accurate rainfall estimates. For example, over Oklahoma, out of twenty-one 1.0º boxes, thirteen met the standard error and signal to noise ratio requirements for reliable comparison. This information for selected boxes has been used in an initial analysis. Results show a seasonal pattern in the 1.0ºX1.0º daily bias estimates. Bias has been computed as the mean satellite estimate divided by the mean surface estimate. The pattern reveals bias maxima in late spring or early summer and again during the winter while the fall season shows that the satellite estimate typically underestimates. The 2.5ºX2.5º monthly comparisons reveal a maximum in absolute error during the summer months. The monthly estimates also tend towards overestimation as seen in the bias values. Results for additional locations will also be shown.