Analysis of regional rainfall-runoff parameters for the Lake Michigan Diversion hydrological modeling
The Lake Michigan Diversion Accounting (LMDA) system has been developed by the U.S. Army Corps of Engineers, Chicago District (USACE-Chicago) and the State of Illinois as a part of the interstate Great Lakes water regulatory program. The diverted Lake Michigan watershed is a 673-square-mile watershed that is comprised of the Chicago River and Calumet River watersheds. They originally drained into Lake Michigan, but now flow to the Mississippi River watershed via three canals constructed in the Chicago area in the early twentieth century. Approximately 393 square miles of the diverted watershed is ungaged, and the runoff from the ungaged portion of the diverted watershed has been estimated by the USACE-Chicago using the Hydrological Simulation Program-FORTRAN (HSPF) program. The accuracy of simulated runoff depends on the accuracy of the parameter set used in the HSPF program. Nine parameter sets comprised of the North Branch, Little Calumet, Des Plaines, Hickory Creek, CSSC, NIPC, 1999, CTE, and 2008 have been developed at different time periods and used by the USACE-Chicago. In this study, the U.S. Geological Survey and the USACE-Chicago collaboratively analyzed the parameter sets using nine gaged watersheds in or adjacent to the diverted watershed to assess the predictive accuracies of selected parameter sets. Six of the parameter sets, comprising North Branch, Hickory Creek, NIPC, 1999, CTE, and 2008, were applied to the nine gaged watersheds for evaluating their simulation accuracy from water years 1996 to 2011. The nine gaged watersheds were modeled by using the three LMDA land-cover types (grass, forest, and hydraulically connected imperviousness) based on the 2006 National Land Cover Database, and the latest meteorological and precipitation data consistent with the current (2014) LMDA modeling framework.
Results indicate that the North Branch and Hickory Creek parameter sets, which belong to the original calibration group, attained an overall “satisfactory” rating on monthly runoff volumes based on the three performance statistics selected, but the annual and over-the-period runoff volumes were generally underestimated. Parameter sets CTE and 2008 attained a similar satisfactory rating on monthly runoff volumes but the annual and over-the-period runoff volumes were overestimated in general. Although the percent bias was improved, the CTE and 2008 parameter sets also had increased residuals in monthly runoff volumes and decreased quality of the model fit to the measured streamflows relative to the North Branch and Hickory Creek parameter sets. The NIPC and 1999 parameter sets, on the other hand, had larger percent bias and residuals in monthly runoff volumes, and underestimated the annual and over-the-period runoff volumes.
Recalibration of the HSPF parameters to the updated inputs and land covers was completed on two representative watershed models selected from the nine by using a manual method (HSPEXP) and an automatic method (PEST). The objective of the recalibration was to develop a regional parameter set that improves the accuracy in runoff volume prediction for the nine study watersheds. Knowledge about flow and watershed characteristics plays a vital role for validating the calibration in both manual and automatic methods. The best performing parameter set was determined by the automatic calibration method on a two-watershed model. Applying this newly determined parameter set to the nine watersheds for runoff volume simulation resulted in “very good” ratings in five watersheds, an improvement as compared to “very good” ratings achieved for three watersheds by the North Branch parameter set.
|USGS Numbered Series
|Analysis of regional rainfall-runoff parameters for the Lake Michigan Diversion hydrological modeling
|Scientific Investigations Report
|U.S. Geological Survey
|Illinois Water Science Center
|vii, 55 p.
|Albers Equal-Area Conic projection
|Online Only (Y/N)
|Additional Online Files (Y/N)
|Google Analytic Metrics