Complexity and uncertainty in hydrological modeling for urban areas of varying database quality.

摘要

The relationship between complexity, data available, uncertainty and error in modeling of urban hydrological systems was investigated using the USEPA (United States Environmental Protection Agency) SWMM/RUNOFF (Stormwater Management Model/Runoff Module). As part of this effort, models that calculate complexity (COMPLES - Complexity Simulator) and parameter uncertainty (UNCES - Uncertainty Simulator) for SWMM/RUNOFF were developed. COMPLES quantifies complexity due to parameters and process aggregation and allows the comparison and classification of SWMM/RUNOFF models according to their complexities. UNCES computes the boundaries of the model output as a function of the uncertainty in the model parameters. It applies Full Factorial (2n and 3n) and Monte Carlo techniques and allows the comparison of the precision of models. Applications of COMPLES and UNCES are presented. Finally, an experiment was developed to establish a relationship between model complexity, available data, uncertainty and error. SWMM users were asked to develop models for a development of approximately 50 ha with different levels of complexity and available data. The complexity and uncertainty of these models were quantified using COMPLES and UNCES. Complexity and data acquisition are shown to reduce model precision and accuracy, respectively.