Specification of Predictors Necessary for the Determination of over or Underestimation of Radar Derived Total Rainfall.
Abstract
A study was performed to produce a set of empirical predictors useful in identifying over or underestimation of radar determined rainfall rates. Two radar sampling procedures, systematic and irregular, were employed against light and heavy rain events to calculate rainfall rates for 10 km x 10 km grid boxes. Sampling intervals used were 10, 15, 20, 30, 40, and 50 minutes. The sampled rainfall rates were compared against the rainfall rates for 5-minute sampled data and the error was determined. The predictors were then use to determine whether the error was an over or underestimation. There was no significant difference in results as a function of the sampling interval used. Different sets of predictors were required for the category of precipitation as well as over or underestimation. The percentage of correct identifications, whether for over or underestimation, approached 80%. Allowing for error associated with the 5-minute sampling interval, the correct decision approached and sometimes exceeded 90%. The predictors which produced the best results utilized percent coverage of the grid box by precipitation, temporal and spatial homogeneity, and the mean rainfall rate.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1987
- Accession Number
- ADA185363
Entities
People
- Vincent P. Holbrook
Organizations
- Air Force Institute of Technology