Mine rehabilitation and vegetation recovery verification
Progressive mine rehabilitation is a legal obligation, but ground inspection is slow and intermittent. Multi-sensor time series from Sentinel-2, Landsat and GEDI lidar can verify genuine vegetation recovery against approved plans, season by season.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 5-day revisit at mid-latitudes with both satellites. Red-edge bands (B5, B6, B7 at 20 m) are particularly sensitive to canopy chlorophyll content, distinguishing stressed re-vegetation from healthy cover.
- Landsat 8/9 OLI: 30 m resolution, 16-day revisit per satellite (8-day combined). Archive extends to 1972 for Landsat 1, giving pre-disturbance baselines that Sentinel-2 cannot provide. SWIR bands resolve bare soil fraction and litter cover alongside green vegetation.
- NASA GEDI: Space-borne full-waveform lidar operating from the ISS. Footprint diameter approximately 25 m, spaced 60 m along-track and 600 m across-track. Measures canopy height and vertical structure directly, which passive optical indices cannot. Sparse sampling means not every rehabilitation polygon will have a usable shot; careful spatial matching is required.
- Planet SuperDove: 3 m resolution, near-daily revisit. Eight spectral bands including red-edge. Useful for resolving fine-scale patchiness within rehabilitation polygons that 10 m imagery blurs, and for confirming whether a green signal is spatially coherent vegetation or artefact.
Why a single inspection visit proves almost nothing
Mine rehabilitation conditions in most jurisdictions require progressive re-vegetation to meet cover thresholds, species diversity targets and canopy height benchmarks before a bond is released. A regulator or independent auditor visiting once a year will almost certainly arrive after rain, when even bare compacted spoil can carry a temporary green flush of annual weeds. That flush can look, from the ground, indistinguishable from genuine native re-vegetation.
Satellite time series removes that ambiguity. A genuine recovering plant community shows a characteristic phenological signature: moderate greenness in dry season, higher greenness in wet season, and a slow year-on-year upward trend in indices like NDVI and fractional cover as canopy closure increases. A rain-flush on bare ground spikes sharply after a rainfall event and collapses within weeks. Distinguishing these two patterns requires dense temporal sampling across multiple seasons, which is exactly what Sentinel-2's five-day revisit provides.
What the indices actually measure, and where they mislead
NDVI (Normalised Difference Vegetation Index) is the workhorse: it uses the ratio of near-infrared to red reflectance and responds to chlorophyll density. For rehabilitation monitoring, fractional cover decomposition is often more informative. Published methods, including the Joint Remote Sensing Research Programme (JRSRP) approach used in Australian mine rehabilitation assessments, decompose each pixel into photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV) and bare soil fractions. NPV, which includes standing dead material and litter, is an important rehabilitation indicator that NDVI alone misses because it does not absorb in the red band.
Sentinel-2's red-edge bands add another dimension. Chlorophyll absorption in the 700-740 nm range is detectable before a canopy is dense enough to produce a strong NDVI signal, which means early-stage re-vegetation can be tracked before it would register clearly in a Landsat scene. The honest limit: at 10-20 m resolution, a rehabilitation polygon smaller than roughly 0.5 ha becomes difficult to assess cleanly because mixed pixels at polygon edges contaminate the spectral signal. Planet SuperDove at 3 m reduces that problem substantially, at the cost of a commercial licence.
Canopy height: the dimension optical sensors cannot see
Cover indices confirm that something green is present. They do not confirm that it is tall enough, or structurally complex enough, to meet rehabilitation targets that specify canopy height or basal area. That is where GEDI contributes. NASA's Global Ecosystem Dynamics Investigation instrument derives relative height metrics (RH50, RH75, RH98) from full-waveform lidar returns, giving direct estimates of canopy height and vertical structure.
The practical constraint is real. GEDI's ground track spacing of approximately 600 m across-track means that a 50 ha rehabilitation polygon may contain only a handful of usable footprints, or none at all if cloud or the ISS orbit geometry produces a gap. Where shots do land, they provide ground-truth quality canopy height data that can calibrate the relationship between Sentinel-2 canopy reflectance and height, extending that relationship spatially across the polygon. Where shots are absent, the analyst must be honest with the client: height estimates are modelled, not directly measured.
Building the rehabilitation trajectory against a reference ecosystem
A target NDVI or fractional cover value is meaningless without context. The standard approach is to identify reference polygons: areas of intact native vegetation of the same ecosystem type, in the same climate zone, that have not been disturbed. Tracking the same indices over the same time series in reference polygons gives the expected seasonal envelope. Rehabilitation polygons are then assessed against that envelope, not against an arbitrary threshold.
Landsat's archive is essential here. For mines that were opened before 2015, Sentinel-2 data does not exist for the pre-disturbance period. Landsat 5 TM and Landsat 7 ETM+ data stretching back to the 1980s can reconstruct the pre-disturbance spectral signature of the site and establish how the reference ecosystem behaves across drought years, wet years and fire cycles. A rehabilitation trajectory that looks impressive in a single wet year may look far less so when placed against a 30-year reference envelope.
Satellize applies this multi-sensor trajectory approach in its analytics work, including the crop-estimation programme for the Kingdom of Tonga, where separating genuine seasonal productivity from transient green response is a structurally similar problem.
Delivering evidence a regulator will accept
The output of a rehabilitation verification programme is not a map. It is a defensible, time-stamped record of how vegetation cover has changed, relative to the approved plan and the reference ecosystem, across every reporting period. That means polygon-level statistics, not raster images: mean and standard deviation of NDVI, PV fraction and NPV fraction per polygon per quarter, with confidence intervals, flagged against the plan's milestones.
Where GEDI shots are available and spatially matched, canopy height estimates are appended. Change alerts, triggered when a polygon's trajectory falls below a statistically defined threshold relative to its reference envelope, can be delivered within days of a Sentinel-2 acquisition. The honest caveat on latency: persistent cloud cover over tropical or sub-tropical mine sites can produce gaps of several weeks in the optical record. SAR coherence can detect large-scale disturbance through cloud, but it does not substitute for optical vegetation indices. Clients in high-cloud environments should expect irregular reporting intervals rather than fixed monthly cadence.
Typical figures
| Spatial resolution (primary) | 10 m (Sentinel-2 visible/NIR), 20 m (Sentinel-2 red-edge/SWIR), 30 m (Landsat OLI) |
| Spatial resolution (high-resolution check) | 3 m (Planet SuperDove, commercial licence required) |
| Revisit frequency | 5 days (Sentinel-2 dual satellite, mid-latitudes); 8 days (Landsat 8+9 combined); near-daily (Planet) |
| Canopy height measurement | GEDI footprint ~25 m diameter; across-track spacing ~600 m; height accuracy ±1-3 m published for closed canopy |
| Spectral bands used | Red, NIR, red-edge (705 nm, 740 nm, 783 nm), SWIR (1610 nm, 2190 nm) for fractional cover decomposition |
| Minimum polygon size for reliable assessment | ~0.5 ha at 10 m resolution; ~0.1 ha at 3 m resolution |
| Archive depth | Sentinel-2: from 2015; Landsat: from 1972 (Landsat 1); GEDI: from April 2019 |
| Cloud limitation | Optical sensors blind under cloud; tropical sites may have gaps of 2-6 weeks in wet season |
| Reporting latency (clear sky) | Polygon statistics typically available within 3-5 days of satellite acquisition |
| Delivery formats | GeoTIFF time-series stacks, polygon-level CSV/Excel statistics, GeoJSON alert layers, PDF audit reports |
Analytics Satellize can run
| NDVI and fractional cover time series per rehabilitation polygon | Spectral index calculation and JRSRP-style fractional cover decomposition on Sentinel-2 and Landsat surface reflectance | Quarterly GIS layer and CSV table with per-polygon PV, NPV and bare-soil fractions |
| Rehabilitation trajectory scoring against approved plan milestones | Comparison of observed fractional cover trajectory against plan thresholds and reference-ecosystem seasonal envelope | Annual compliance report with polygon-level pass/flag/fail status and supporting time-series charts |
| Rain-flush versus genuine re-vegetation discrimination | Temporal shape analysis of NDVI time series: spike-and-collapse pattern versus sustained upward trend across multiple seasons | Flagged alert layer identifying polygons where green signal is likely transient, delivered within 5 days of triggering acquisition |
| Reference ecosystem spectral envelope | Multi-decadal Landsat archive analysis of intact native vegetation polygons in same climate zone | Reference envelope dataset (mean ± 1 SD by month) used as the baseline for all trajectory scoring |
| Canopy height validation layer | Spatial matching of GEDI RH98 shots to rehabilitation polygons; height-to-reflectance model where shot density permits | GeoJSON point layer of GEDI-derived canopy height estimates with shot-count and uncertainty flags per polygon |
| Pre-disturbance baseline reconstruction | Landsat TM/ETM+/OLI archive analysis to characterise site vegetation before mining commenced | Historical NDVI and cover statistics report for use in bond-release assessments and legal proceedings |
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.