How to tell a tannery from a textile mill in satellite imagery

When a candidate-site list comes back from a regional scan, half the work is just sorting the hits. A processing facility shows up as a cluster of roofs, a yard, a water feature, and nothing else tells you what's actually happening inside until you learn to read the layout. Tanneries and textile mills get confused constantly in this first pass, and the mix-up matters: miscategorize a site and you've either wasted a field visit or missed the tier-2 leather supplier your tier-1 garment maker never mentioned.

The water feature is usually the fastest tell

Both industries are wet processes and both sit near a river or a drainage canal for a reason. But the effluent ponds look different.

Tannery effluent, especially from chrome tanning, often carries a persistent blue-green or turquoise cast in the holding ponds and settling basins. It's consistent week to week because the chemistry is the same batch after batch. If you're scanning the same site across a few passes and that pond color barely shifts, that's a chrome tanning signature, not a dye house.

Textile mill effluent, by contrast, changes color with the dye lot. One month the discharge pond might read dark red, the next month navy, the next pale yellow. Inconsistent hue across time is more consistent with a dye and finishing mill than a tannery. If you only have one image, this test doesn't work. It needs a short time series, which is one reason annual revisits of the same region matter more than a single snapshot.

Yard activity and roofline tell you the rest

Tanneries handle raw hides, and raw hides need space. Look for open-air drying yards: long rows of frames, racks, or stretched hides laid flat, sometimes salt piles near the receiving dock where trucks offload. The main production building tends to be low, squat, and added onto in pieces over the years, because tanning happens in a sequence of drums and pits that gets built out as the operation grows rather than designed as one unit.

Textile mills read differently from the roof down. Weaving and knitting sheds often still use the old saw-tooth, north-light roof pattern, visible from above as a repeating sawtooth texture across a single large building. Spinning mills tend to be taller, boxier, multistory structures instead of sprawling single-story sheds. The yard holds bales of raw cotton or yarn under cover rather than hides laid out to dry, and you'll often see a dedicated bale or cone storage building separate from the main process floor.

Boiler stacks show up at both, so a chimney alone doesn't settle it. What settles it is the combination: stack plus sawtooth roof plus bale yard points mill; stack plus drying frames plus hide delivery apron points tannery.

Where visual ID stops being enough

None of this replaces ground truth. A roofline and a pond color get you a confident guess, not a confirmed classification, and a site that looks like a tannery from the air could be a half-converted facility doing something else entirely by the time anyone drives out to check. The imagery's job is narrowing a region down to the handful of sites worth that trip, not closing the file on its own.

That's the actual use case behind building a candidate-site list from a region's physical signatures in the first place: you're not trying to prove who's in the building, you're trying to find the buildings nobody declared so your team knows where to point the follow-up. The visual signatures above are a starting filter for that list, and they get more reliable the more passes of the same site you can compare.

If your supplier map stops at tier 1, this is the kind of detail work it's built to surface.