Wet cooling vs dry cooling: how each shows up in thermal satellite imagery

A gas or coal plant's cooling system tells you almost as much about what's running as the stack does. The trouble is that wet cooling and dry cooling leave very different fingerprints in a thermal band, and if you're pattern-matching across plants without knowing which cooling design you're looking at, you'll misread the imagery.

Wet cooling: the plume you can see

Wet cooling towers (natural draft or mechanical draft) work by evaporating water, and that shows up two ways. On a cool morning, the visible water vapor plume rising off the tower is obvious in plain RGB, no thermal band required. But vapor plumes come and go with ambient humidity and temperature, so they're a weak signal on their own. The more reliable tell is the cooling water itself: a discharge canal, outfall, or return line running back to a river, lake, or cooling pond will run warmer than the receiving water around it. In a thermal band that temperature differential is a clean, durable signature, visible well after any vapor plume has dissipated and largely indifferent to whether the tower is natural or mechanical draft.

The practical read: more units online, more heat rejected, a wider and hotter discharge plume at the outfall. Plants on once-through cooling (pulling straight from a river or estuary with no tower at all) show this same discharge signature without any tower structure to confuse it with. For a wet-cooled plant, the discharge plume is usually the more stable of the two signals week to week, which matters if you're trying to build a consistent nowcast series rather than chase a vapor cloud that depends on the weather that morning.

Dry cooling: no plume, different signature

Air-cooled condensers (ACCs) skip water entirely. Steam from the turbine goes straight to large fin-tube banks arranged on an elevated deck, usually the big A-frame or delta structure you'll recognize immediately once you've seen one. No tower, no vapor plume, no discharge canal to track. That's the whole point of dry cooling in water-constrained sites, but it also removes the signal analysts lean on for wet-cooled plants.

What's left is the ACC deck itself. The fin banks run warmer than ambient when the plant is rejecting heat, and that temperature rise across the deck is visible in a thermal band even though there's no plume or discharge to confirm it visually in RGB. The signature is subtler than a discharge plume: it's a temperature gradient across a structure rather than a hot patch of water against a cold background, and it needs a clean baseline read on the deck at idle to compare against. Stack activity becomes the corroborating signal here, since the ACC deck alone gives you heat rejection but not combustion.

Why this matters for a weekly read

Scan a fleet without checking each plant's cooling configuration first and you'll mix up these two readings fast. A dry-cooled plant will never give you a discharge plume, no matter how hard it's running, so the absence of one tells you nothing about dispatch. A wet-cooled plant on a humid, still morning might show a modest vapor cloud even near minimum load, while the discharge canal tells the real story. Confirm the cooling type for each facility once, then weight the stack, discharge, and ACC-deck signals for that specific plant instead of applying one rule across the fleet.

This is exactly the kind of plant-specific read that's hard to do by eye on a one-off image and easier to do as a running series, which is the idea behind a weekly imagery-derived activity signal you can chart against generation and fuel data for a named plant rather than waiting on the next disclosure cycle.

If you're tracking a specific coal or gas plant and want its stack, discharge, and cooling signatures turned into a consistent weekly read, that's what Power Plant Activity is built for.