Evaluating the impacts of urbanization and stormwater management practices on stream response

Abstract

Land use changes related to urbanization can have profound impacts on runoff characteristics, resulting in changes to the magnitude, frequency, and duration of flow in stream channels. These flow regime changes often alter the quality of aquatic habitats and native biota of streams. Identification of linkages between urban land use policies and practices and the associated geomorphic and ecological consequences in urban streams is needed when evaluating the effectiveness of urban stormwater runoff management practices, including the management of urban development and limiting percent of impervious surface cover. In this study, protocols were developed to help standardize data generation for identifying mechanistic linkages to provide public agencies with tools to identify management practices that will achieve the fewest ecological impacts and sustain physical habitats and ecological conditions in urban streams. The protocols were applied in the North Carolina Piedmont. Eight watersheds spanning a gradient of urbanization from undeveloped to highly developed were used to identify linkages between indicators of aquatic ecosystem health and hydrologic and geomorphic metrics derived from a 20-year continuous stream flow record. The 20-year continuous stream flow records for each watershed were generated using EPA Storm Water Management (SWMM5) models calibrated to 18 months of measured flow data. The macroinvertebrate metrics Ephemeroptera, Plecoptera, and Trichoptera (EPT) richness, EPT percent richness, and a benthic index of biotic integrity (B-IBI) for the eight North Carolina Piedmont watersheds were found to be linearly related to the T0.5 hydrologic metric computed for 1-, 2-, 5-, 10-, and 20-year time periods. Qualitative multihabitat EPT richness metric values were strongly related to the T0.5 hydrologic metric, which describes the percent of time discharge is above the peak discharge of the 0.5-year return interval storm. The T0.5 values computed from the 1- and 2-, and 10-year time periods were the best predictors of macroinvertebrate metric scores. T0.5 values computed from both 15-minute and hourly time steps were good predictors of macroinvertebrate metric values; T0.5 values computed from daily time steps were poor predictors of macroinvertebrate response. The TQmean hydrologic metric (percent of time discharge is above the mean annual discharge value) was not a good predictor of macroinvertebrate response, and the T1 hydrologic metric (percent of time discharge is above the peak discharge of the 1-year return interval storm) did not predict macroinvertebrate response as well as the T0.5 hydrologic metric. The T2 hydrologic metric (percent of time discharge is above the peak discharge of the 2-year return interval storm) was a better predictor of macroinvertebrate response than both the T0.5 and the T1 hydrologic metrics. T2 values computed from an annual maximum series, rather than a partial duration series, were better predictors of macroinvertebrate response for the longer (10- and 20-year) time periods, but offered approximately the same level of prediction when computed for the 5-year time period. Macroinvertebrate metrics were shown to be logarithmically related to predominant substrate size, with EPT richness for the richest targeted habitat having the strongest correlation. Predominant grain size was better predictor of macroinvertebrate metrics than cumulative excess shear stress, except for EPT richness for the qualitative multi-habitat results. Similar relationships were observed between the macroinvertebrate metrics and the cumulative excess shear stress computed for the 1-, 2-, 5-, 10-, and 20-year time periods. The T0.5 hydrologic metric was affected by both land use and channel geometry. When land use in a watershed was changed from rural to urban, but the channel geometry remained the same, the value of the T0.5 hydrologic metric did not decrease as much as when both land use and channel geometry were changed. When the channel geometry of the streams modeled for the undeveloped Morgan Creek watershed were changed to be similar to those observed in urban watersheds of the North Carolina Piedmont, but land use remained the same (rural), the T0.5 hydrologic metric was nearly the same as the one computed when both land use and channel geometry were changed. This indicates that channel geometry plays a significant role in values of the T0.5 hydrologic metric. Modification of rural channel cross-sections to be similar to cross-sections observed in urban watersheds of similar size changed the T0.5 hydrologic metric enough to cause it to predict a degraded EPT richness score from good-fair or fair, to poor. When stormwater controls were applied to the developed Morgan Creek watershed, and when channel geometry was left in its original state, changes in the value of the T0.5 hydrologic metric were minimized, and anticipated changes to macroinvertebrate ratings were reduced. When stormwater controls were applied to a developed Morgan Creek watershed, but channel geometry was changed to be similar to cross-sections observed in urban watersheds, the T0.5 hydrologic metric score was greatly improved, but was not improved enough to minimize changes in the macroinvertebrate score, which was almost always reduced to fair. Long-term changes in erosion potential in the Morgan Creek watershed were minimized when stormwater controls were applied. Biotic response was shown to be related to both hydrologic or geomorphic metric values, indicating that restoration of hydrology may not allow for the attainment of biotic goals if habitat is not also restored. In urbanizing situations, preservation of a more natural hydrology and restrictions on increases in erosion potential will reduce geomorphic changes from what would occur if stormwater management controls were not implemented, although changes in upland sediment supply also must be accounted for. Control of channel erosion was shown to be important because channel geometry was found to directly influence hydrologic metric values due to its impact on conveyance capabilities and loss of flood attenuation due to disconnectivity from floodplains. These relationships were shown for the North Carolina Piedmont, but are also applicable throughout the United States.