Case study · worked example

Dalli-Rajhara open-pit mines & Hitkassa tailings dam

A worked example over a working iron-ore mine complex in Balod, Chhattisgarh, India. We used only public satellite images. There was no site visit and no data from the operator. Prepared 25 Oct 2026. We picked this site because it has both steep pit walls and a tailings dam, the two things the method most needs to handle well.

Status: not yet checked on the ground

This is not an engineering assessment, and it says nothing about whether the site is safe. The numbers below are movements measured from satellite, with their uncertainty stated. They need checking against the operator's own survey data.

Download the 2-page summary (PDF)

Key numbers

What we looked at

51

satellite passes, 11 Jan 2025 – 16 Oct 2026

21

months of movement history

40 m

map detail: one reading per 40 × 40 m square

~35–40

mm a year, moving away from the satellite, at two areas

Map of how fast the ground moved over Dalli-Rajhara, with markers for the stable reference point, Hitkassa tailings dam, main pits, Rajhara Hill workings W and pit wall E. — How fast the ground moved, Jan 2025 – Oct 2026. Red means moving away from the satellite, which usually means sinking or sliding downhill. Blue means moving toward it. Blank areas gave no reliable reading. Background: Esri World Imagery. Click for full size.

Chart of movement over time at five locations, Jan 2025 to Oct 2026. Rajhara Hill workings W and pit wall E move steadily to about -60 mm. The stable reference point stays near zero. — Movement over time at five locations, measured against a stable point and the first date. The sharp drop in July–August 2025 at all the mine points is most likely monsoon weather affecting the signal, not real movement.

Results

How much each location moved

Negative numbers mean moving away from the satellite. All movement is measured against a stable point about 5–6 km from the mine.

Location	mm a year	Total mm, 21 months	What it means
Rajhara Hill workings W 20.566°N 81.053°E	−40	−64	Check Clearly above background noise, steady in both years
Pit wall E 20.569°N 81.104°E	−36	−58	Check Clearly above background noise
Main Dalli-Rajhara pits 20.569°N 81.092°E	−11	−18	Too small Can't be told apart from noise
Hitkassa tailings dam area*	−16	−27	Too small No reading on the dam wall itself
Stable reference point 20.617°N 81.109°E	−0.8	−0.3	Stable Zero by definition

Background noise: on ground far from the mine that should be still, readings scatter by about ±8 mm a year, and most fall between −16 and +11 mm a year. So only movement faster than about 16–20 mm a year clearly stands out.
*The satellite got no reliable reading on the Hitkassa dam wall itself. The value shown is the average of 19 reliable points within 500 m, so it does not describe the dam crest. Overall, 34% of the study area and 49% of the mining lease gave reliable readings.

Conclusion: two small areas on the Rajhara Hill workings have moved steadily, about 35–40 mm a year. They are worth checking against the mine's survey data. This run says nothing reliable about the Hitkassa dam.

Limits

What this study cannot tell you

Not checked on the ground. We had no survey prisms, piezometers, GPS or slope-radar data to compare against.
Only one viewing angle. The satellite sees this site from one direction only, so we can't separate up-down movement from sideways movement. Pit walls often move mostly sideways.
Mining itself can look like movement. Digging, dumping and haul-road work change the surface. A "moving" area may be material that was moved, not unstable ground.
No reading on the Hitkassa dam wall. At this level of detail the dam wall gives no reliable signal. Small metal reflectors fixed to the wall, or sharper processing, would fix this.
Some noise remains. Moisture in the air, mainly during the monsoon, still affects some readings. On stable ground the scatter is about 8 mm a year. We did not apply the extra weather corrections a full project would use.
A second satellite did not help yet. We compared with NASA and ISRO's new NISAR satellite for Jun–Oct 2026. Its early data is still provisional, has a gap from Jan to Jun 2026 and did not agree with our main results, so we did not use it.

Limits of satellite measurement in general

It is not real-time. There is one reading per satellite pass, every 6–12 days. It shows trends; it cannot warn of a collapse in progress. Ground-based slope radar is the tool for that.
It measures one direction. It sees movement toward or away from the satellite. Splitting that into up-down and sideways needs views from two directions.
Some ground gives no reading. Thick vegetation, water, snow and freshly dug or dumped ground lose the signal. Small reflectors can fill critical gaps.
It needs checking. Compare results with your survey, prism or GPS data before acting on them.

Next steps

What a full project would add

Sharper maps (20 m), plus point-by-point readings on dam crests and structures
Weather corrections from global weather data, and extra quality checks
A second viewing direction, for example from NISAR, to split movement into up-down and east-west
Low-cost metal reflectors on embankments where the satellite gets no reading
Checks against the operator's survey, prism or GPS data
A longer movement history, back to 2016, and an automatic alert after each satellite pass, with limits agreed with the geotechnical team

Get a free history of your site

Technical details, for remote-sensing specialists

Radar data: Copernicus Sentinel-1, C-band (5.5 cm), relative orbit 92 descending, bursts 092_196415_IW2 + 092_196416_IW2
Interferograms: 147 pairs (each date with the next 3), NASA/ASF HyP3, ISCE2 multi-burst, 40 m pixels
Time series: MintPy 1.6.4 small-baseline inversion; troposphere corrected by phase–elevation correlation; DEM-error correction; pixels with temporal coherence < 0.7 masked
Reference: Coherent ground at 20.617°N 81.109°E
Area of interest: 80.99–81.13°E, 20.53–20.62°N
Noise: Formal velocity σ ≈ 1.5 mm/yr; empirical 1σ ≈ 7.9 mm/yr on coherent ground > 3 km from the workings; no ERA5/GACOS correction or unwrapping-error correction; phase-closure ambiguities in parts of the workings
Cross-check: NISAR L-band GUNW (provisional), Jun–Oct 2026, ionosphere-corrected; 20–50 mm residual ramps, not used

Contains modified Copernicus Sentinel data 2025–2026 processed by ASF HyP3 (NASA). NISAR data courtesy NASA/ISRO via ASF DAAC (provisional). Basemap: Esri World Imagery. Software: ISCE2 (via HyP3), MintPy (Yunjun et al., 2019).