Overcoming Obstacles in Applying SWMM to Large-Scale Watersheds
The RUNOFF module of the U.S. EPA’s Storm Water Management Model (SWMM) was used to simulate the watershed hydrology and water quality contaminant loadings for the Schuylkill River Basin. The Schuylkill River is over 130 mi. (210 km) long, includes over 180 higher-order tributaries draining more than 2,000 square mi. (5,100 square km). For the model of this basin, simulated runoff loadings were accumulated at major junctures, or inlet points, for up to 14 land use categories in each of the 356 sub-watersheds, resulting in over 3,000 RUNOFF module sub-basins. This allows results to be summarized by land use, by model basin, by accumulated groupings of sub-watersheds, and for the entire watershed study area. Continuous SWMM was applied using a 15-min simulation time step, with full implementation of the snowmelt and groundwater subroutines, for execution periods of 30 y, to generate estimates of seasonal and annual watershed discharge and loadings.
We believe this to be one of the largest applications of SWMM RUNOFF in terms of the combination of the size of the drainage area, the number of sub-basins, and the number of simulation time steps. The use of SWMM at this scale uncovered numerous limitations of the model requiring either modifications to the SWMM code or the implementation of functional work-arounds. For example, array bounds set for the input and simulation data required modifications to the SWMM code expanding the limits. Also, during the calibration of the model, SWMM did not allow retrieval of daily flows at select inlet points, if those points are routed to downstream inlets. Also, the limit on the scratch file size restricted the length of simulations and essentially prevented the use of other SWMM modules in this modeling exercise. Furthermore, the addition of water quality constituents significantly increases the size of the scratch files, further limiting the simulation period, which eventually was reduced to one year. Our resolution of these issues included reducing simulation periods and external processing using SAS.
The limitations decreased the efficiency of modeling with SWMM at these scales and have led us to begin a process to make improvements in SWMM to better facilitate the large-scale applications required to support current regulatory programs in the United States. Our goal is to improve SWMM to accommodate fully featured hydrology and water quality applications for watersheds at scales in excess of 10,000 mi2 (25,000 km2) using 5,000 to 10,000 RUNOFF sub-basins for simulation periods of 30 y or more at 15-min time steps.
This paper is only available in PDF Format:
View full text PDF