Multi-depth suspended sediment estimation using high-resolution remote-sensing UAV in Maumee River, Ohio

Satellite remote-sensing has been widely used to map suspended sediment concentration (SSC) in waterbodies. Current development of the unmanned aerial vehicle (UAV) technology allows mapping of SSC at finer spatial resolution providing high flexibility in terms of cost and acquisition time. However, the technology is immature and transfer of empirical algorithms from existing remote-sensing technologies to UAV still has to be explored. This study uses the MicaSense Sequoia sensor with four bands (green, red, red edge, and near-infrared [NIR]) mounted on-board a fixed-wing UAV to map SSC within the Maumee River in Ohio, USA, at multiple depth intervals (15, 61, 91, and 182 cm). The simple linear and stepwise regression models show the advantage of multiple bands and band ratios over single bands in mapping SSC. The findings show a limited performance of the Sequoia sensor when compared to field spectroradiometer measurements. In all cases but one, the adjusted coefficient of determination ((Formula presented.)) values is lower for the UAV data. The regression equations become similar at and below a depth of 0–61 cm, and (Formula presented.) become constant at and below a depth of 0–91 cm. While the spectroradiometer-related equations are sensitive to a wider spectral range (from green at the surface to NIR wavelength at 182 cm depth), the UAV-related equations are insensitive to green spectrum and they include a narrower spectral range (from red to NIR) over all depth increments. Field spectroradiometer measurements exhibit a strong relationship with cumulative SSC at 182 cm depth (0–182 cm) ((Formula presented.)) whereas UAV reflectance data show the best relationship with SSC at 91 cm (0–91 cm) ((Formula presented.)) suggesting that ~91 cm may be an optimal depth for UAV under given conditions. The results show that UAVs can be a practical but somewhat limited tool to monitor SSC in small- to medium-sized rivers.