Tailings Pond Water Depths from Space : A Comparison of Mine Water Depth Measurement Methods

This paper demonstrates the potential of satellite-based methods as a complementary tool for surveying tailings ponds in active deposition tailings impoundments. Three analyses were performed to compare tailings pond bathymetry from single-beam echosounder surveys (herein referred to as physical surveys) with satellite-derived bathymetry (SDB). The depths of the ponds from physical survey data ranged from 0.4 ft to 28 ft. Two of the SDB results were developed using a physics-based method that employs a Radiative Transfer Model (RTM). An additional SDB was calculated using a machine-learning approach. For all analyses, three-meter (3-m) resolution bathymetry was derived from 8-band PlanetScope satellite images acquired concurrently with the physical surveys. Two sites were evaluated using RTM-based evaluations. The available physical survey for each site helped guide the RTM calculations, but the data was not directly used in any algorithms or analyses. In direct comparisons of physical surveys with SDB data, the RTM methodology for SDB produced depths that were, on average, 2.4 ft deeper than physical survey data at both sites. These depth differences resulted in reclaim pond volumes that were 27% and 40% greater than the corresponding physical survey estimates. One additional SDB evaluation was performed using a random forest machine-learning approach. The machine-learning model was directly trained on available physical survey data. As expected, this approach was better at matching physical survey depths; however, the analysis's results are not likely to be transferable to other sites, nor may they translate as effectively as the RTM methodology for changes over time or for applying lessons learned to future site locations. Volumes were not estimated for this analysis. To reach parity with physical surveys, SDB analysis requires a combination of investigations, including using higher-resolution satellite imagery with improved radiometric quality and potential changes to fundamental methodologies and practices. Additional investigations and integrated collection methods could optimize bathymetric data acquisition and volumetric assessment to benefit operations planning and environmental monitoring of pond sites.