Simulation of flood hydrographs for Georgia streams

Abstract

Flood hydrographs are needed for the design of many highway drainage structures and embankments. A method for simulating these flood hydrographs at urban and rural ungauged sites in Georgia is presented. The O'Donnell method was used to compute unit hydrographs from 355 flood events from 80 stations. An average unit hydrograph and an average lag time were computed for each station. These average unit hydrographs were transformed to unit hydrographs having durations of one-fourth, one-third, one-half, and three-fourths lag time and then reduced to dimensionless terms by dividing the time by lag time and the discharge by peak discharge. Hydrographs were simulated for these 355 flood events and their widths were compared with the widths of the observed hydrographs at 50 and 75 percent of peak flow. The dimensionless hydrograph based on one-half lag-time duration provided the best fit of the observed data. Multiple-regression analysis was used to define relations between lag time and certain physical basin characteristics, of which drainage area and slope were significant for the rural equations, with impervious area being added for the Atlanta urban equation. A hydrograph can be simulated from the dimensionless hydrograph, peak discharge of a specific recurrence interval, and lag time obtained from regression equations for any site of less than 500 sq mi in Georgia. For simulating hydrographs at sites larger than 500 sq mi, the U.S. Geological Survey computer model CONROUT can be used. CONROUT produces a simulated outflow discharge hydrograph with a peak discharge of a specific recurrence interval. The diffusion analogy routing method with single linearization was used in this study.