Case Study of the Chicago River Watershed: Physical Modeling vs Data-driven Modeling of an Urban Watershed
StatisticsView Usage Statistics
We developed a water quality model for the highly urbanized Chicago River watershed based on hydrologic simulation using BASINS/HSPF. Appropriate consideration was given to the effective impervious area (EIA). The 5 y water quality simulation resulted in finding total nitrates loadings at both point and nonpoint sources. However, it is always useful to have modeling alternatives to validate the simulation results of a physically based model with a data-driven one. Data-driven modeling has gained a lot of attention in recent decades in both hydrology and water resources research. While physically based models require the description of system inputs, physical laws and boundary and initial conditions, a data-driven model simply extracts knowledge from a large amount of data with only a limited number of assumptions about the physical behaviour of the system. For this case study, both data-driven and physical models were considered to simulate total nitrates. Comparing the performance of the two modeling approaches, the data-driven models show better performance. RMSE for regression models showed an increase in prediction performance of up to 10.7 %. Data-driven models require fewer inputs and can be deployed anywhere in the watershed, while physical models require extensive data inputs and can only be applied to the specific watershed outlets selected in the simulation. These arguments suggest the complementary use of both physical and data-driven models. The physical model can be a planning tool whenever significant physical change takes place in the watershed. The data-driven model can be an operating tool that can be periodically used to inspect the watershed water quality parameters, especially if TMDL and WQS are established for the watershed.
|Journal Title||Journal of Water Management Modeling|
|Rights||In Copyright (All Rights Reserved)|