Thermal evaluation of urbanization using a hybrid approach

Urban development increases runoff temperatures from buildings and pavement, which can be harmful to aquatic life. However, our ability to predict runoff temperature as a function of land use is limited. This paper explores available tools for simulating runoff temperature with respect to brook trout (Salvelinus sp.), a sensitive species. The Minnesota Urban Heat Export Tool (MINUHET) and the Storm Water Management Model (SWMM) were applied to a 14.1 km² portion of the Stroubles Creek watershed near Blacksburg, Virginia for two summers. Streamflow, water temperature, and weather data were acquired from the Virginia Tech StREAM Lab (Stream Research, Education, and Management) monitoring stations. SWMM and MINUHET were calibrated and validated for streamflow, and stream temperature, respectively. The models were sensitive to imperviousness (SWMM-predicted streamflow) and dew point temperature (MINUHET-predicted water temperature). While the models output time-step was 15 min, the model performance in simulating streamflow was evaluated using Nash-Sutcliffe Efficiency (NSE) on hourly time-steps. NSE values were 0.67 and 0.65 for SWMM and 0.62 and 0.57 for MINUHET during the calibration and validation periods, respectively, indicating that SWMM performed better than MINUHET in streamflow simulation. Stream temperatures were simulated using MINUHET with NSE value of 0.58 for the validation period, demonstrating a satisfactory simulation of water temperature. Since SWMM is not capable of stream temperature simulation beyond simple mixing. Hydrologic and thermal outputs from SWMM and MINUHET were combined in a hybrid approach that emphasized the strength of each respective model, i.e. SWMM for runoff and streamflow and MINUHET for water temperature. Heat loads were simulated using the MINUHET and the Hybrid models; the Hybrid model (0.56) had a greater NSE than MINUHET (0.45) alone. MINUHET predictions indicated water temperatures would exceed the trout toxicity threshold of 21 °C during 39% and 38% of calibration and validation periods, respectively. Since the observed temperature exceeded the toxicity threshold 59% and 53% of the time for the calibration and validation periods, respectively, MINUHET was not a conservative predictor of the duration of temperatures exceeding the toxicity threshold value.