Relative information from thermal infrared imagery via unoccupied aerial vehicle informs simulations and spatially-distributed assessments of stream temperature.

Abstract

Stream temperature is a measure of water quality that reflects the balance of atmospheric heat exchange at the air-water interface and gains or losses of water along a stream reach. In urban areas, stormwater sewers deliver water with varying magnitude and temperature to streams at variable timescales. Understanding the impacts of stormwater through space and time is therefore difficult to do with conventional approaches like in situ sensors. To study the impacts of stormwater on creek water temperatures, we combined in situ water temperature observations with thermal infrared (TIR) imagery collected via unoccupied aerial vehicle (UAV). Imagery was collected in May, June, and July of 2017. As ongoing work with UAV-based TIR suggests that this imagery is prone to poor accuracy, we focused on creating several data products beyond absolute water temperatures that can be used to assess temporal and spatial water temperature variations. In particular, TIR data products were used to extract the length of the observed stormwater plume as well as the width of the creek cross-section impacted by stormwater. From these values, we conclude that relatively narrow stormwater plumes affecting a small fraction of creek width can alter creek water temperatures for considerable distances downstream. We also applied TIR data to constrain results of a deterministic stream temperature model (HFLUX 3.0) that simulates the physical processes affecting stream heat exchanges. Stormwater plume lengths obtained from TIR imagery were used to refine spatially-distributed simulations, demonstrating that relative temperature information obtained from UAV imagery can provide useful calibration targets for stream temperature models. Overall, our work demonstrates the added value of UAV datasets for understanding urban stream temperatures, calibrating water quality models, and for modeling and monitoring of the impact of spatially explicit hydrologic processes on stream temperature.