Calibration of Runoff Curve Numbers and Rational Runoff Coefficients for the Kansas Metropolitan Area

Abstract

The runoff curve number (CN) and the rational runoff coefficient (C) for undeveloped land are key inputs to common methods for estimating flood quantiles. This document describes the calibration of these inputs with Kansas City-area streamflow data and recommends changes to the current Section 5600 storm-drainage design criteria for the Kansas City area. Runoff curve numbers and rational runoff coefficients for undeveloped land were calibrated to give the best possible estimates of flood quantiles. The curve numbers and runoff coefficients were considered frequency-dependent, so separate calibrations were performed for annual exceedance probabilities of 50%, 20%, 10%, 4%, 2% and 1%. The calibrations were performed on 28 gaged rural watersheds in the Kansas City area. In aggregate, these watersheds have physical characteristics that are similar to those of undeveloped land and urban open space in the Kansas City area. Basin lag times and times of concentration were computed with the new calibrated equations for the Kansas City area developed by the University of Kansas (McEnroe et al., 2015). Rainfall frequency estimates from NOAA Atlas 14 Volume 8 (2013) were used in all calibrations. Runoff curve numbers were calibrated for use in the Baseline Unit Hydrograph Method in Section 5600. Using generalized least-squares regression, regional flood-frequency equations were developed from the peak-flow records for the 28 gaged watersheds. The curve numbers were calibrated so that the Baseline UH Method yields the same peak flows as regional flood-frequency equations. A second set of curve-number calibrations were performed using HEC frequency storms of 24-hour duration in place of the 24-hour NRCS Type 2 storms specified in Section 5600. Rational runoff coefficients for each frequency of interest were calibrated by fitting the rational equation to the log-transformed data for the 28 gaged watersheds by least-squares regression. On average, the calibrated curve numbers for the NRCS 24-hr Type 2 storm and the HEC storm of 24-hr duration fall below the CN value of 74 used in Section 5600 for undeveloped land and urban green space. Calibrated curve numbers decrease slightly with decreasing annual exceedance probability. The calibrated rational runoff coefficients are strongly frequency-dependent and generally higher than the values recommended for undeveloped land in the Section 5600 design criteria.