Date of Award
Fall 10-25-2023
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Civil, Construction, and Environmental Engineering
First Advisor
Walter McDonald
Second Advisor
Anthony Parolari
Third Advisor
Somesh Roy
Abstract
The hydrologic urban heat island occurs when thermally polluted stormwater runoff is exported from hot urban centers to fragile downstream ecosystems. Mitigating runoff with higher temperatures requires an understanding of the relationship between land surface temperatures (LST) and stormwater runoff; however, LST is difficult to define due to the complexity of heterogenous urban catchments. Therefore, this thesis combines in-situ monitoring with thermal imaging via small unmanned aerial systems (sUAS) to understand how urban catchment heterogeneity influences runoff temperature. Based upon this data, a framework for modeling heat flux in stormwater runoff using sUAS LST datasets is presented. Monitoring was conducted in Milwaukee, WI from July to November 2022, capturing 12 storm events across four urban catchments: greenspace (2,200 m2), parking lot (1,800 m2), roadway (1,100 m2), and roof (175 m2). Monitoring data included continuous temperature measurements at the outlet of each catchment, as well as LST of the catchments captured before and after storm events with a thermal camera on a DJI M100 sUAS. Using this data, a distributed process-based modelling framework was developed to simulate the heat flux in stormwater runoff. Results from monitoring indicate that the catchments exhibited distinct runoff and land surface temperatures. For example, there were clear distinctions between impervious surfaces on the ground and those of the building. The roof catchment was among the highest median LSTs (34.1 °C), but had the lowest event mean temperatures (EMT) (19.8 °C); comparatively, the roadway’s median LST was 1.7 °C cooler than the roof, yet its median EMT was 2.1 °C hotter. The roof’s cool EMTs are likely due to rapid LST changes (20 °C per hour) and a cool air-conditioned subsurface. Additionally, LST variability among land covers was wide, exhibited by standard deviations: canopy cover (0.68 °C), concrete (1.44 °C), shrub/mulch (1.36 °C), grass (0.68 °C), asphalt (1.62 °C), and roof (1.37 °C). This variation indicates the potential error of generalizing LSTs. The proposed modeling framework was applied to an example storm and demonstrates the use of a raster foundation to represent the spatial distribution of LST in the simulation of temperature in stormwater runoff.