DEVELOPMENT AND UTILITY OF HUFF CURVES FOR DISAGGREGATING PRECIPITATION AMOUNTS

Abstract

Watershed models are becoming more sophisticated and require temporal inputs of precipitation to drive themodeled hydrologic processes. Short time increments of the order of minutes are particularly needed, but these are seldomavailable or do not have good spatial coverage. The most widely available precipitation data in the United States are totalamounts for 24-h periods. Information on the actual intensity distribution within a period of precipitation is lacking andpractitioners resort to approximate methods to distribute the precipitation in time (disaggregation). The most commonapproach to disaggregate total precipitation amounts has been to use design storms, fixed patterns of the time distributionof precipitation intensities within a period. Huff curves provide a method of characterizing storm mass curves. They area probabilistic representation of accumulated storm depths for corresponding accumulated storm durations expressed indimensionless form. The development of Huff curves is described in the present study using precipitation data from Coshocton,Ohio because the procedure has never been documented in the literature. The potential use of these curves, the state ofknowledge, and research needs for advancing the utility of Huff curves for storm disaggregation are summarized. Threeapproaches for using Huff curves to disaggregate precipitation totals are described: design storms (fixed patterns ofintensities), stochastic simulation of within-storm intensities, and a hybrid of these two approaches. Expanded use of Huffcurves as described in the latter two approaches offer opportunities for maximizing the information contained in Huff curves.There is no meaningful correspondence between NRCS design storms (Types I, IA, II, and III) and Huff curves. A computerprogram is available to facilitate the development of Huff curves for stimulating research into their practical use.