| Summary: | Much of the geomorphic work of rivers occurs underwater. As a result, high resolution<br />quantification of geomorphic change in these submerged areas is important. Currently, to quantify this<br />change, multiple methods are required to get high resolution data for both the exposed and submerged<br />areas. Remote sensing methods are often limited to the exposed areas due to the challenges imposed<br />by the water, and those remote sensing methods for below the water surface require the collection of<br />extensive calibration data in-channel, which is time-consuming, labour-intensive, and sometimes<br />prohibitive in dicult-to-access areas. Within this paper, we pioneer a novel approach for quantifying<br />above- and below-water geomorphic change using Structure-from-Motion photogrammetry and<br />investigate the implications of water surface elevations, refraction correction measures, and the<br />spatial variability of topographic errors. We use two epochs of imagery from a site on the River Teme,<br />Herefordshire, UK, collected using a remotely piloted aircraft system (RPAS) and processed using<br />Structure-from-Motion (SfM) photogrammetry. For the first time, we show that: (1) Quantification of<br />submerged geomorphic change to levels of accuracy commensurate with exposed areas is possible<br />without the need for calibration data or a dierent method from exposed areas; (2) there is minimal<br />dierence in results produced by dierent refraction correction procedures using predominantly<br />nadir imagery (small angle vs. multi-view), allowing users a choice of software packages/processing<br />complexity; (3) improvements to our estimations of water surface elevations are critical for accurate<br />topographic estimation in submerged areas and can reduce mean elevation error by up to 73%;<br />and (4) we can use machine learning, in the form of multiple linear regressions, and a Gaussian Naïve<br />Bayes classifier, based on the relationship between error and 11 independent variables, to generate a<br />high resolution, spatially continuous model of geomorphic change in submerged areas, constrained by<br />spatially variable error estimates. Our multiple regression model is capable of explaining up to 54%<br />of magnitude and direction of topographic error, with accuracies of less than 0.04 m. With on-going<br />testing and improvements, this machine learning approach has potential for routine application in<br />spatially variable error estimation within the RPAS−SfM workflow.
|
|---|
