An Empirical Study of Flood Measurement Error

Abstract

At a gaging station, very large peak discharges are usually estimated by extending the rating curve or by making an indirect measurement; hence these discharges are subject to much larger errors than are directly measured discharges. A crude measure of the uncertainty of the largest observed peak discharge at a gaging station is the ratio of the largest directly measured discharge to the largest observed peak discharge. For 46 gaging stations in the State of Wisconsin this ratio ranges from 0.16 to 0.99, with a median of 0.72. For the annual flood series from the half of the gaging stations with the lowest ratios, the coefficients of skewness have considerably higher mean and standard deviation than do those of the remaining flood series. These results are consistent with the impact of measurement error, though they can also be explained by the occurrence of single extremely large floods in several of the annual flood series.