Design and implementation of hydrologic unit watersheds for rainfall-runoff modeling in urban areas.

摘要

The calibration of complex hydrology and hydraulics of rainfall-runoff models represents one of the most challenging problems in water resources engineering. Unlike undeveloped watersheds, but specifically urban basins with surface drainage. From the available models, SWMM (Storm Water Management Model) was used as the modeling engine since it was developed for urban watersheds.;Calibration procedure used a Multi-Criteria Decision Analysis (MCDA) approach that minimized the RMSE (Root Mean Square Error) between the flow duration curves of the modeled and the observed runoff. The flow duration curve was divided in High and Low Flows using the 1-Yr storm to split the curve, since there is a change in flow regime at this point. Pareto optimal front surfaces were obtained.;Two case studies in North Carolina (Pigeon and SW Prong basins) were used to illustrate a proposed methodology for calibration. The methodology simplified the drainage network and irregular sub-catchments shapes were converted to regular shapes using a Kinematic Wave (KW) cascading plane approach.;The KW cascading plane approach showed to be effective to convert irregular sub-basins shapes to rectangular features. A discretization analysis was performed where a set of hydrologic experiments using different levels of discretization were used and a threshold discretization value in urban hydrology was investigated. Needed GIS data was extracted through a toolbox. MCDA methodology and numerical simulations showed that Horton's decay coefficient (K, 1/h) and drying time (Tw, days) needed to have different values for the High and Low Flow portions of the flow duration curve to improve performance. Longer drying times were required to improve estimation of High Flows than Low Flows because the soils would take more time to recover their initial infiltration capacity.;The Representative Element Area (REA) concept was explored in SWMM and it was found that sub-catchment sizes of 3% of the total basin size were appropriate. This magnitude represents the suggested level of discretization in urban watersheds since the improvement in performance became asymptotic either to 1.00 (Pearson's Moment Correlation Coefficient---PMCC, Nash-Sutcliff Coefficient---NSC and Index of Agreement---IOA) or to zero (RMSE) and therefore, it is not significant to improve the spatial resolution. Coarser resolution levels underestimated peak flow rates and total runoff volumes. Research results are summarized in a proposed protocol to discretisize urban watersheds.