Abstract
The spatial variability of surface rainfall over 5- and 30-day time periods is observed, and it is found that the spatial decorrelation length of precipitation is comparable to the size of a single surface gauge network. The observed variability is found to affect radar-derived precipitation estimation, particularly if it is based on calibration using rain gauges. The radar subgrid-scale variability is also observed using some redundant and finer-scale gauge networks deployed during the Tropical Rainfall Measuring Mission (TRMM) ground-validation field campaigns. Based upon statistical analysis and a point-based decision-making system, a best-suited spatial–temporal filtering technique is suggested and, when applied to match radar data with any other surface observation, is found to reduce bias.
Corresponding author address: Dr. Saswati Datta, Joint Center for Earth Systems Technology (JCET), University of Maryland, Baltimore County, ACIV Wing A, Rm 114, 1000 Hilltop Circle, Baltimore, MD 21250. sdatta@umbc.edu