Potash mine tailings pile volume estimation for environmental liability assessment

Repeat-pass SAR and optical stereo photogrammetry track potash tailings pile volume and lateral spread independently of operator reporting, converting elevation-change rates into production proxies and environmental-liability indicators for ESG debt covenants and mine-closure bond sizing.

Sensors

TanDEM-X (bistatic X-band SAR pair): Generates single-pass interferometric DEMs at 12 m posting with published relative vertical accuracy of approximately 2 m over moderate slopes; repeat acquisitions allow differencing to detect volume change between epochs. Commercial tasking required; archive coverage patchy before 2010.
Sentinel-1 (C-band SAR, ESA): Free, 6-day repeat at mid-latitudes (12-day for a single satellite); 5 × 20 m IW mode resolution. Suitable for InSAR-derived subsidence and heap-edge displacement mapping, though vertical uncertainty over steep, salt-saturated slopes rises to 2–4 m and temporal decorrelation on loose salt surfaces can be severe.
Pleiades / Pleiades Neo (optical stereo): Tri-stereo or along-track stereo at 30–50 cm native resolution; photogrammetric DSM accuracy of 0.3–1 m vertically under cloud-free conditions. Provides the highest-fidelity volume snapshots but is blocked by cloud and requires two tasked passes within hours. Archive stereo pairs exist for some major potash districts from 2012 onward.
ICEYE (X-band SAR microsatellites): Spotlight mode delivers 1 m resolution imagery with same-day revisit possible through constellation scheduling. Useful for detecting lateral heap-edge movement at sub-weekly cadence and for monitoring brine pond extent adjacent to tailings. DEM generation requires coherent repeat passes, which the constellation is increasingly able to provide.

What a salt heap reveals that a balance sheet does not

Potash extraction produces roughly one tonne of salt tailings for every tonne of product shipped. The heaps are enormous, stable in shape over weeks, and almost impossible to disguise. A mine that reports reduced output but whose tailings pile continues to grow at a consistent rate is sending a contradictory signal. Conversely, a pile that stops growing during a period of claimed maintenance shutdown confirms the shutdown. This physical consistency is what makes remote sensing useful to lenders, insurers and regulators who cannot rely solely on operator disclosure.

The financial stakes are specific. Mine-closure bonds in several jurisdictions are sized against the volume of waste requiring remediation. If the declared volume is understated, the bond is undersized and the liability falls to the state or, in insolvency, to creditors. ESG-linked loan covenants increasingly reference independently verified environmental metrics. Satellite-derived volume estimates, with their uncertainty ranges stated honestly, give counterparties a defensible independent data point.

How elevation differencing turns radar into a volume gauge

The core method is straightforward in principle and demanding in practice. A digital surface model (DSM) is generated at time T1, another at time T2, and the two are co-registered and subtracted. Positive residuals indicate material added; negative residuals indicate removal or compaction. Integrating the residual surface over the heap footprint gives a volume change figure. Multiply by bulk density, typically 1.1–1.3 tonnes per cubic metre for compacted salt tailings, and you have a mass estimate.

TanDEM-X bistatic acquisition is the most reliable SAR route because both antennas image simultaneously, eliminating atmospheric phase delay differences between passes. Sentinel-1 repeat-pass InSAR works for subsidence monitoring and edge-displacement mapping but is more vulnerable to temporal decorrelation on loose, hygroscopic salt surfaces. Optical stereo from Pleiades gives better vertical precision but only when cloud cover cooperates. In practice, a programme that combines Sentinel-1 for frequent change detection with occasional Pleiades stereo for absolute calibration outperforms either sensor used alone.

Vertical uncertainty is not a footnote; it is the analysis

SAR-derived DEMs carry vertical uncertainties of 1–3 m depending on slope angle, surface moisture and baseline geometry. For a heap that grows 2 m per year, a 2 m vertical uncertainty means the annual volume-change confidence interval can span zero. This is not a reason to abandon the method; it is a reason to be explicit about it. A properly propagated uncertainty budget converts a point estimate into a range, and a range is still actionable if the lower bound exceeds the threshold that triggers a covenant review.

Slope matters disproportionately. Heap flanks steeper than roughly 20 degrees amplify layover and foreshortening artefacts in SAR geometry, pushing local vertical errors toward the upper end of the published range. Pleiades stereo is less sensitive to slope but introduces its own errors from image matching failures on homogeneous white salt surfaces. Analysts should report which surfaces are within reliable accuracy bounds and which are not, rather than applying a single global uncertainty figure to the whole heap.

Brine pond expansion as a corroborating signal

Tailings heaps are rarely the only observable. Potash mines generate large volumes of saturated brine, stored in evaporation ponds adjacent to the heaps. Pond area, turbidity and spectral signature change with throughput. Sentinel-2 multispectral imagery at 10 m resolution can track pond extent and approximate brine concentration through colour change in the visible and near-infrared bands. A heap growing in volume while adjacent ponds simultaneously expand is a stronger production signal than either observation alone.

SAR backscatter from brine surfaces is also distinctive: very smooth saline water returns a low specular signal that contrasts sharply with rougher surrounding terrain. This makes pond-edge detection reliable even under partial cloud, where optical coverage fails. Combining SAR-derived pond extent with DSM-derived heap volume gives a two-variable check that is harder to reconcile with false reporting.

Translating volume change into liability and production proxies

The conversion from cubic metres to financial exposure requires a chain of assumptions, each of which should be stated. Bulk density of compacted halite tailings is reasonably well constrained at 1.1–1.3 t/m³ from published geotechnical literature on potash waste. The ratio of tailings volume to product volume depends on ore grade, which varies by deposit. Published figures for major potash districts in Saskatchewan, Germany and Belarus suggest tailings-to-product ratios of roughly 3:1 to 5:1 by mass, but these should be treated as order-of-magnitude guides rather than precise calibrations unless site-specific data are available.

For closure-bond sizing, the relevant output is total heap volume at assessment date, not the change rate. For production monitoring, the change rate over a defined period is what matters. For ESG covenant compliance, both may be relevant: a covenant might specify that heap volume must not grow beyond a permitted level, or that remediation of a designated section must be demonstrable by a given date. Each use case calls for a slightly different analytic output, and the uncertainty budget must be propagated appropriately for each.

Satellize structures these analytics as periodic GIS deliverables with explicit confidence intervals, calibrated against available Pleiades stereo epochs. The Tonga crop-estimation programme demonstrated a similar principle of pairing a high-accuracy reference observation with lower-cost repeat monitoring; the same architecture applies here.

Practical limits and what to do about them

Cloud cover over temperate potash districts, particularly in Germany and the UK, can interrupt optical stereo acquisition for weeks at a time. SAR fills most of this gap but cannot fully substitute for optical stereo in absolute volume calibration. A programme that relies exclusively on Sentinel-1 InSAR without periodic optical ground-truth will accumulate drift in its volume estimates over multi-year periods.

Heap compaction is a genuine ambiguity. A surface that descends by 0.5 m between epochs could represent material removal or simple gravitational settling of loose salt. Compaction rates for potash tailings are not well characterised in the open literature, and analysts should flag this as an unresolved source of systematic bias in any volume-change report. Where mine operators publish compaction monitoring data, incorporating it into the model reduces this uncertainty materially. Where they do not, the honest position is to bound the estimate rather than resolve it artificially.

Typical figures

Best achievable vertical accuracy (optical stereo, cloud-free)	0.3–1 m (Pleiades / Pleiades Neo tri-stereo)
Typical SAR DEM vertical accuracy	1–3 m (TanDEM-X); 2–4 m (Sentinel-1 InSAR, slope-dependent)
Spatial resolution for heap-edge mapping	1 m (ICEYE Spotlight); 5 × 20 m (Sentinel-1 IW); 30 cm (Pleiades)
Repeat interval	6 days (Sentinel-1 constellation); same-day possible (ICEYE); on-demand (TanDEM-X, Pleiades)
Minimum detectable volume change	Approximately 50,000–100,000 m³ per epoch at SAR accuracy; lower with optical stereo calibration
Brine pond extent detection	10 m (Sentinel-2 optical); SAR backscatter pond-edge at 5–20 m depending on sensor
Archive depth	Sentinel-1 from 2014; TanDEM-X global baseline from 2010–2015; Pleiades stereo archive from 2012 (site-dependent)
Delivery formats	GeoTIFF DSM differencing rasters, GeoJSON heap-footprint polygons, CSV volume-change time series with confidence intervals, PDF assessment report
Cloud impact	SAR unaffected; optical stereo blocked; temperate sites may lose 30–60% of optical acquisition windows seasonally

Analytics Satellize can run

Heap volume and volume-change time series	DSM differencing from repeat TanDEM-X or Pleiades stereo acquisitions, co-registered to a common datum	Quarterly GeoTIFF differencing rasters and CSV time series with propagated vertical uncertainty bounds
Heap footprint and lateral-spread polygon	SAR backscatter edge detection and optical supervised classification of heap extent	GeoJSON polygon layer updated per acquisition, with area and perimeter statistics
Production-proxy index	Volume-change rate converted to mass-change using published bulk-density range for halite tailings; normalised to a baseline period	Monthly index value with uncertainty range, delivered as structured JSON feed or PDF summary
Brine pond extent and change detection	Sentinel-2 NDWI and SAR specular-return thresholding for pond-edge delineation	Monthly pond-area polygon and area-change statistic, flagged if pond expands beyond a client-defined threshold
Closure-bond liability volume assessment	Absolute DSM from Pleiades stereo referenced to pre-mining terrain model; heap volume integrated over footprint polygon	Point-in-time assessment report with volume estimate, confidence interval, and comparison to operator-declared figure
Covenant-trigger alert	Rule-based threshold applied to volume-change or footprint-expansion metrics derived from above analytics	Email or API alert when monitored parameter crosses a pre-agreed threshold, with supporting imagery and data extract

Sources

Who does the work

We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.