Beyond volume proxies: An exposure-based scenario-screening framework for conditional potential loss of life in tailings storage facility portfolios

Tailings storage facilities (TSFs) are expanding, and occasional failures can cause severe downstream impacts. This study presents a screening-level framework for conditional Potential Loss of Life (PLL, given failure) using open datasets, Google Earth Engine (GEE), and an Excel release model. The workflow estimates dam geometry by DEM cut-and-fill, caps release at 0.90 of storage, routes a MERIT Hydro-guided corridor, applies exponential thickness decay with minimum thickness trimming, and evaluates PLL on the WorldPop 100 m grid.
Each site is evaluated using 144 equally weighted deterministic combinations and summarised with scenario-exceedance curves, scenario-set Value-at-Risk at 95% (VaR95) and scenario-set Expected Shortfall at 95% (ES95). Across 17 sites, upper-tail PLL spans nearly three orders of magnitude, with VaR95 from 0.1 to 148 and P99 from 0.15 to nearly 200. A high-consequence group emerges, while most remain low. Scenario-exceedance curves reveal varied distributions, including long-tail behaviour where typical outcomes are low but some combinations produce severe PLL. Partial rank analysis showed a stronger independent association of PLL with exposed population (ρpartial = 0.64) than with stored volume (ρpartial = 0.16), although the 17-site sample limits generalisation.
The framework supports standardised ranking and prioritisation, complementing site-specific breach hydraulics and evacuation analyses. Results should be interpreted within assumptions on routing, release estimation, transferable depth-fatality curves, population uncertainty and equal design weighting. Estimates represent conditional consequences, not probabilistic annual risk. Future work should extend beyond PLL, which captures only fatality consequences and cannot substitute for assessing displacement, livelihoods, environmental harm or wider social consequences.