Floods - How Much is Too Much Water
Authors: Thomas Over (USGS) , Riki Saito (USGS) , David Soong (USGS) , T-Y Su (USACE-Chicago District)
Trends in streamflow records are often seen as a potential signature of climate change, but land-use change may be the dominant driver of trends in some settings. Analyzing temporal trends with respect to precipitation and urbanization allows their relative effects to be estimated. In this study, the annual peak flow data series during the period from 1945-2009 from 143 USGS streamgaging stations in the Northeastern Illinois (Chicago metropolitan) region with drainage areas less than 200 sq. mi. were analyzed by multiple linear regression to jointly estimate the effects of urbanization and precipitation on their trends. Urbanization was expressed as the urbanized fraction of each basin and was estimated by using an adjusted decadal Census-based measure of housing density. Precipitation was estimated as the maximum daily rainfall during the period three days before and through one day after the date of the peak, distributed by using Thiessen polygons. It was found that changes in the logarithm of peak flows can be modeled as directly proportional to changes in urbanization and precipitation, with proportionality constants of about 0.51 and 0.085, respectively. The precipitation coefficient may be under-estimated as a result of uncertainty in the precipitation estimates, but it was retained in the regression model to improve the accuracy of the estimated urbanization effect. Without precipitation in the regression, the urbanization coefficient increased to about 0.53, indicating that without precipitation in the analysis, the small increasing trend in precipitation is assigned erroneously to the effects of urbanization. No strong evidence was found region-wide that urbanization in later decades of the analysis period had a smaller impact on peaks than in the earlier decades, despite institution of stormwater detention requirements at varying times beginning in 1972 throughout the region. Using these results, the peak flow data series will be adjusted to present (2010) urbanization conditions to create data that are stationary with respect to urbanization (though not precipitation changes) for use in updating regional regression equations for estimating flood peak quantiles in urbanized areas of northeastern Illinois.
Keywords:
How to Cite: Over, T. , Saito, R. , Soong, D. & Su, T. (2014) “Identifying and Adjusting for Effects of Urbanization on Peak Streamflows”, National Hydraulic Engineering Conference 2014. 1(2014).