Alluvial Gold in the Indus, Gilgit and Hunza Rivers: How We Actually Assess It

I spent three days last September squatting next to a sluice on a Gilgit River terrace, arguing with a panner about whether his recovery was 60% or closer to 40%. He insisted 60. My test pan said 38.

That argument is basically the whole story of placer gold assessment in this part of Pakistan. Everyone has a number. Almost nobody has a method behind the number.

So let me walk through how we actually do resource assessment on alluvial gold Pakistan deposits — specifically across the Indus, Gilgit and Hunza river systems where GBX holds ground. This is for the buyers and JV partners who keep asking us for a JORC-style summary and deserve to know what's behind it.

Why these three rivers matter geologically

The gold in the Indus, Gilgit and Hunza isn't a mystery. It's eroded out of the Kohistan Island Arc, the Karakoram batholith, and the auriferous shear zones along the Main Karakoram Thrust and the Main Mantle Thrust. Hundreds of kilometres of bedrock source, grinding into glacial meltwater, dumping into braided river systems that have been working as natural gravity concentrators for — conservatively — the last two million years.

That's a long time to make a placer.

The Hunza river gold is typically finer and flakier. Lots of flour gold, which is exactly the fraction local panners lose. The Indus, especially below the Gilgit confluence near Jaglot and downstream toward Chilas, carries coarser material and the occasional nugget — we've personally weighed a 4.7 gram piece pulled from a bar near Thakot. Gilgit river material sits between the two in grain size distribution.

Different grain size means different recovery equipment. Which means if you assess all three systems with one methodology, you'll get bad numbers. I did this wrong on our first internal estimate back in 2021 and had to redo six months of work.

The methodology, step by step

Here's how we run an assessment on a placer block now. It's not exotic. It's the Wells/Clarkson approach adapted for braided Himalayan rivers, with some adjustments we learned the hard way.

1. Terrace and bar mapping first, sampling second.

Before anyone touches a shovel, we map the active channel, the low terraces (T1, T2), and the high terraces (T3+) using a combination of Sentinel-2 imagery and ground walks. Active bars get re-graded every monsoon, so they're a renewable but lower-grade target. High terraces are where the historical concentration sits — old channel positions, sometimes 15 to 40 metres above present river level.

We've found values in T2 gravels near Bunji that ran 380 mg/m³. Active channel bars in the same stretch averaged closer to 95 mg/m³. Big difference in economics.

2. Bulk sample, not pan sample.

For placer gold Indus assessment, a gold pan tells you presence/absence. It does not tell you grade. We dig test pits — typically 1m × 1m × depth-to-bedrock or depth-to-water — and process the entire excavated volume through a calibrated sluice with known recovery characteristics.

Minimum bulk sample we'll accept for a grade estimate: 0.5 cubic metres. Below that the nugget effect destroys you.

3. Recovery calibration with salted tests.

This is the step most local operators skip. We salt our sluice runs with weighed gold of known grain size distribution (we use crushed and sized material from a previous bulk concentrate) and measure what comes out the other end. Our current field sluice runs about 72% recovery on +200 mesh material and drops to maybe 31% on −200 mesh.

If you don't know your recovery, your grade number is fiction.

4. Strip ratio and overburden logging.

A lot of the better terrace gravels sit under 2 to 6 metres of barren or low-grade silt and aeolian cover. Honest resource statements have to include strip ratio. We log every test pit with overburden thickness, gravel thickness, gold-bearing horizon thickness, and bedrock character (where reached).

5. Block modelling with conservative kriging.

We break the licence area into 50m × 50m × 1m blocks and interpolate grade between sample points using inverse distance weighted to start, then ordinary kriging once we have enough samples for a variogram. For placers, the variogram ranges are short — typically 80 to 200 metres along channel — so you need denser sampling than people expect.

What we won't claim

Honestly, I get tired of the WhatsApp messages claiming "5 grams per tonne placer" in the Indus. That's not how placer grades work and it's not what's in these rivers.

Realistic average grades, based on our work to date:

• Active channel bars, Gilgit and Hunza systems: 70 to 140 mg/m³
• Low terrace gravels (T1-T2), favoured sections: 150 to 400 mg/m³
• High terrace pay streaks, where we've found them: 400 to 900 mg/m³, but discontinuous

At $2,600/oz gold and current Pakistani diesel prices, the economic cutoff sits somewhere around 110-130 mg/m³ for a mechanised operation, lower if you're running gravity-only with no chemicals.

The tonnage story is more interesting than the grade story. The cumulative volume of unconsolidated gravel in just the stretch from Jaglot to Chilas runs into hundreds of millions of cubic metres. Even at modest grades, the contained metal is substantial. The question is always: how much can you actually mine without picking a fight with the river, the monsoon, or the Gilgit-Baltistan EPA?

Licensing reality

Placer operations in GB fall under the Gilgit-Baltistan Mines & Minerals Department. Reconnaissance licences are straightforward. Mining leases for alluvial operations require an approved feasibility, an EIA cleared by GB-EPA, and increasingly, community benefit agreements with the local union council. We've found the community piece is the one foreign partners underestimate the most.

If you want to look at the actual sample data on a specific block — Bunji, Jaglot, or our Hunza concession near Chalt — write to me directly. I'd rather send you the pit logs than try to summarise them here.

What's the recovery on your sluice, by the way? Because if you don't know, neither of us knows what's in the ground.