| Stormwater runoff data for flow and water quality were collected for three years from seven locations in five urban karst watersheds ranging in size from 93.5 to 231.7 hectares (total impervious areas ranging from 23.4% to 67.3%). The measured data were then used to develop, calibrate (with targets of 30% for peak runoff rate and 10% for runoff volume), and validate a SWMM for each karst watershed. The hydrologic response, water quality characteristics, and model parameters were compared among the study watersheds with different land uses and percentages of impervious areas and then the response, characteristics, and parameters from the karst watersheds were compared to values from non-karst watersheds. Recharge areas and drainage pathways dominated the hydrologic response in the watersheds where these karst features have been preserved.;Median runoff ratios (RR, runoff volume/rainfall volume) for these urban karst watersheds (0.02 to 0.28) were found to be significantly less than regional values ranging from 0.4 to 0.5 and less than values from the EPA National Urban Runoff Program (NURP) studies (values of 0.2 to 0.62) for watersheds with similar percent imperviousness. Using the RR based on percentage of impervious area from the NURP data to predict runoff would result in overestimating the volume of runoff in these watersheds by 164% to 1,400% in these karst watersheds.;The median event mean concentrations for all constituents except for nitrate that were part of the study described herein and in the NURP studies (total suspended solids, total phosphorus, soluble phosphorus, copper, lead and zinc), were smaller from the study karst watersheds than the lower limit of the range of values from the NURP data.;The commonly referenced model parameter values used in stormwater model development were not adequate or appropriate for these karst watersheds.;The calibrated model parameters of curve number, depth of impervious storage, and depth of pervious storage were outside of the standard range of model parameters. If using an uncalibrated model parameterized with typical reference values, the volume of runoff and peak rate of runoff for these karst watersheds could be over-predicted by over 200% and over 500%, respectively. |
