Leveraging multi-sensor remote sensing for monitoring moisture in tailings storage facilities

The EU-Horizon-funded project Multiscale Observation Services for Mining-related Deposits (MOSMIN) aims to develop comprehensive services for the geotechnical and environmental monitoring of mine waste to improve the safety, efficiency, and transparency of mining operations. Monitoring moisture and water ponding in tailings storage facilities (TSF) is important for safeguarding the geotechnical stability of these structures and optimising operational efficiency, including discharge planning, water reuse, and dust control. 

Satellite-based moisture monitoring of tailings is a cost-effective alternative to traditionally employed in situ sensors, as it provides large-scale and long-term coverage. Spectral indices from multispectral satellite data have been used successfully as moisture proxies in TSF, and as a basis to track size and bathymetry of decant ponds. While this is a viable methodology in arid regions, synthetic aperture radar (SAR), an active technique employing microwaves that can penetrate clouds, provides an alternative in more challenging atmospheric conditions. In this contribution, we explore the possibilities of using Sentinel-1 SAR data for moisture estimation within TSF, focusing on the Talabre TSF in Chile, one of MOSMIN's pilot sites, which, due to its desert climate provides the opportunity for cross-validation with optical soil moisture proxies. A qualitative estimate of relative soil moisture, showing strong agreement with optical-derived moisture estimates, can be obtained by normalising backscatter values using a time series to establish minimum and maximum bounds. However, in contrast to natural soils, which tend to have consistent surface roughness characteristics over time, tailings at Talabre exhibit smooth surfaces when wet and are prone to rapid desiccation, creating rougher surfaces as they dry, which leads to a negative relationship between SAR backscatter and moisture. 

Salts forming on tailings are another complicating factor, acting as barriers to evaporation and obscuring the moisture-related backscatter behaviour outlined above. In SAR backscatter time-series, we observe that certain areas show a slower rate of increase in backscatter during drying periods than others. The spectral signatures extracted from Enmap hyperspectral satellite data suggest that these regions correspond to tailings impacted by salt formation, providing an effective means to identify salt-affected areas (and excluding them from moisture estimation). 

For water body delimitation, thresholding the backscatter intensity based on the premise that water acts as a specular reflector, works well in most situations. For windy areas, we present a new method to delineate water bodies which uses paired Sentinel-1 images from different incidence angles, acquired from different look directions, to detect deviations in backscatter caused by wind-induced Bragg scattering. 

These findings underscore the potential of integrating earth observation data to provide reliable and versatile moisture monitoring services in TSFs for proactive mine management and sustainable resource development.