Unmanned aerial vehicle-based monitoring of groundwater inputs to surface waters using an economical thermal infrared camera

Abstract

Groundwater discharge into streams helps maintain flows during droughts and provides refugia for thermally sensitive species, as the relatively constant temperature of spring water buffers against instream diurnal and seasonal fluctuations. Unmanned aerial vehicle (UAV)-based thermal imagery represents an attractive option to monitor sites of groundwater-surface water mixing. However, the use of economical thermal sensors presents several challenges to obtaining a stable dataset for mosaicking. Here, we present a method for acquiring and postprocessing thermal infrared imagery at spring discharge sites using an inexpensive, uncooled microbolometer mounted on a small UAV. The procedure involves initial estimation and removal of pixel bias from the sensor output, and then compensation for temporal sensor drift by optimizing the stability of the signal at ground features detected within multiple frames. We illustrate this approach by presenting a case study at the Devils River, a groundwater dependent stream in Texas. Comparison with imagery acquired using a more expensive thermal camera system, designed to compensate for sensor drift at the time of data acquisition using knowledge of internal camera temperature, reveals that our method produces a more consistent final mosaic image. A good linear fit (r2 = 0.97) between the signal in the stabilized dataset and ground-based measurements of water temperature underscores the potential for this method to inexpensively produce high quality maps of surface temperature in ecologically important stream reaches.