ABSTRACT
A major source of error in hydrologic models is the poor quantification of the areal distribution of rainfall, typically due to the low density of rain gauges. A rain gauge located at a single point may not represent an extensive area, with only one value. Conceptual rainfall-runoff (CRR) models simulate runoff generation by a variety of conceptual parameters and route the runoff using unit hydrographs to an outlet. CRR models are inherently nonphysics based and lump parameters at the basin or subbasin level. The complex interaction of input with drainage network presents challenges to the design of storm-water drainage infrastructure, the management of flooding, flood mitigation, and real-time forecasting of multiscale urban drainage systems with multiscale inputs. The generalized likelihood uncertainty estimation (GLUE) based prediction limits the capture of uncertainly in model output associated with uncertainly in model parameterization. GLUE provides a useful modeling approach for advancing beyond globally optimized, unique, parameter sets.
