What a 30-Meter Pixel Can Tell You About a Mineral Deposit

Last March I was looking at an SRTM tile of a valley near Skardu. Something off about it. A creek was making a 73-degree bend that no creek should make naturally — water doesn't turn that sharp unless something underneath is forcing it to.

That something was a fault. And along that fault, two of my field teams later confirmed quartz veining with visible sulfides.

This is the part of mineral exploration most people skip. They jump straight to spectral data — Sentinel-2 band ratios, ASTER alteration mapping, the shiny stuff. But terrain came first. Terrain is the skeleton. Everything else is skin.

Why a Free Dataset From 2000 Still Beats Most Field Surveys

SRTM (Shuttle Radar Topography Mission) flew for 11 days in February 2000. NASA and the NGA bounced radar off the planet and built a digital elevation model covering everything between 60°N and 56°S. Pakistan included. All of it. Free.

The resolution most of us use is 30 meters per pixel. That's not amazing by modern standards (we also pull in 12.5m ALOS PALSAR DEM when we need finer detail), but here's the thing — for structural geology, 30m is plenty. You're not looking for a boulder. You're looking for a 4-kilometer lineament that hints at a deep crustal break.

And those breaks are where minerals concentrate. Hydrothermal fluids don't move through solid rock. They move through fractures. Fractures show up as terrain features.

So when someone asks me why we spend so much time on SRTM DEM mining workflows when we have access to fancier data — that's the answer. The terrain tells you where to look before you waste a single rupee on spectral analysis.

What We Actually Look For in the DEM

A few specific things, in roughly the order our pipeline processes them:

Lineaments. Long, straight, or gently curving features that cut across topography. Could be a fault, a dike, a fracture zone, or a lithological contact. We run automated lineament extraction (Hough transform plus some custom filtering Sufyan and the team built to throw out roads, canals, and field boundaries — early on we got fooled by an irrigation channel in Punjab and I won't pretend that wasn't embarrassing).

Drainage anomalies. Rivers and streams are honest. They follow gravity and weakness. When a drainage network suddenly turns 90 degrees, or when streams run parallel for kilometers in an area that should have dendritic patterns, something structural is going on underneath. Trellis drainage in folded terrain. Rectangular drainage over fracture sets. These are textbook clues that still get ignored.

Slope breaks. Sudden changes in slope angle often mark lithological boundaries. A resistant quartzite ridge against a softer schist. A gabbro intrusion punching through limestone. Slope and aspect rasters derived from the DEM make these jump out.

Circular features. Honestly these are my favorite. Ring-shaped topography can indicate intrusive bodies — granites, porphyries, sometimes kimberlites. Porphyry copper systems in Balochistan? They tend to sit under or near these circular signatures. The Reko Diq region shows several on DEM analysis if you know how to filter for them.

Curvature. Profile and planform curvature highlight subtle ridges and valleys the human eye misses. We compute these as standard layers in every project.

None of this is new geology. Structural geologists have been reading topography since the 1800s. What's different now is we can do it across 50,000 square kilometers in an afternoon instead of three field seasons.

A Real Example From Chitral

Client came to us in 2023 with a 240 sq km lease and a budget that wouldn't cover even basic ground geophysics across that area. Classic problem.

We pulled SRTM, generated lineament density maps, drainage anomaly overlays, and slope-aspect derivatives. Cross-referenced with regional geological maps from GSP. The DEM mineral targeting workflow narrowed his lease down to 18 sq km of high-priority ground — basically three structural intersection zones where two or more lineament sets crossed.

Then we layered Sentinel-2 and ASTER on top of just those 18 sq km. Found alteration signatures consistent with mesothermal gold systems in two of the three zones.

He sent a small team. They came back with samples assaying between 2.1 and 6.4 g/t Au from rock chips. Not a deposit yet — that takes drilling — but a target worth drilling, which is the whole point of digital elevation model exploration work. You're not finding the gold from space. You're finding the place that's worth looking for the gold.

I used to think the spectral data was the star of the show. Spent way too much time on band ratios early in my career. Then I realized the structural framework decides whether the spectral signature even matters. A nice clay alteration anomaly in the middle of an undeformed shale basin? Probably nothing. The same anomaly sitting on a lineament intersection at a slope break? Now we're talking.

A Few Practical Things If You're Trying This Yourself

Use void-filled SRTM (the v3 release). The original had data gaps in steep terrain — exactly the terrain you care about for mineral exploration in places like Gilgit-Baltistan or the Sulaiman Range.

Don't trust automated lineament extraction blindly. Always overlay roads, canals, and political boundaries. About 30% of what an unfiltered algorithm calls a lineament is actually human infrastructure.

Combine SRTM with ALOS PALSAR for steep mountain work. The radar penetrates better and the 12.5m resolution catches narrower features.

And please — ground truth. Always. We've had targets that looked perfect on screen and turned out to be a glacial moraine. The mountain has opinions the satellite can't always read.

So the question I keep coming back to with every new project at geomines: before you spend money on spectral analysis, drilling, or boots on the ground — have you actually read what the terrain is already telling you for free?