What is a minimum detection limit in satellite methane monitoring?

If you've spent any time comparing methane monitoring vendors, you've run into the term "minimum detection limit" (MDL) and probably gotten three different answers about what it means. That's not an accident. MDL gets used loosely in sales decks, and the loose usage hides the one question that matters to an MRV analyst: will this sensor actually see the leak I'm worried about, or will it just see the big ones everyone already knows about?

The short definition

Minimum detection limit is the smallest methane emission rate, usually expressed in kilograms per hour, that a given sensor can reliably pick out of the background noise at a specific pixel size and under specific conditions. Below that rate, the plume's SWIR absorption signature is too faint to separate from surface reflectance variation, water vapor interference, and sensor noise. It's still there. The instrument just can't confirm it.

The phrase "reliably pick out" is doing a lot of work in that sentence, and it's where most of the confusion starts. MDL isn't a single fixed number for a sensor the way a thermometer has a fixed resolution. It shifts with:

• Ground sample distance (GSD). A tighter GSD concentrates the same plume over fewer, more saturated pixels, which generally improves sensitivity to smaller sources.
• Surface albedo. Dark, uniform surfaces like a tank pad or bare soil give a cleaner signal than bright sand, water, or snow.
• Wind speed. Higher wind disperses the plume faster, which can push a real emission below the detection threshold even though the source rate hasn't dropped at all.
• Solar geometry and atmospheric conditions. Low sun angle, haze, and cloud edge effects all degrade the signal-to-noise ratio the retrieval algorithm has to work with.

So when a vendor quotes a single MDL figure, ask what wind speed and surface type that number assumes. A detection limit measured over flat, dark terrain at 2 m/s wind doesn't transfer to a pad in sagebrush at 6 m/s.

Why this matters more than coverage area

A lot of satellite methane marketing leads with basin-wide coverage and daily revisit, and those numbers are real and useful. But coverage tells you the sensor flew over the area. It doesn't tell you whether it would have caught the emission you're trying to catch.

This is the gap between area-flux instruments built for basin-level trends and imaging spectrometers built to resolve an individual plume down to a site boundary. The first kind can tell a regulator or an investor that a basin's aggregate emissions went up. The second kind, and the reason point-source detection exists as its own category, is what tells an MRV analyst which pad. If your compliance reporting or your fund's emissions screening needs site-level attribution, the question isn't "does this sensor cover my area," it's "is my typical emission rate above or below this sensor's MDL at my site's surface conditions."

That's the practical test. Pull your facility's historical leak rates, or your fund's portfolio's typical release sizes, and compare them against the detection range a vendor can support for a pixel at your GSD. If most of your events sit near or below that line, you're going to get a lot of false negatives dressed up as clean compliance.

What a lower MDL buys you

A lower minimum detection limit means the sensor can confirm a smaller plume as real rather than noise. For an operator, that's the difference between catching a wellhead valve leak early and only finding out about it when a regional advisory flags your basin weeks later. For a fund analyst screening a portfolio company, it's the difference between seeing an isolated small release and seeing nothing until the emission grows large enough to show up in coarser data, by which point the exposure has already been sitting there unmanaged.

None of this means lower MDL is free. Finer GSD and hyperspectral retrieval generally mean narrower swath width and more tasking constraints, which is why sensitivity and cadence trade against each other on most platforms. The honest way to evaluate a sensor is to ask what MDL it holds at the revisit frequency you need, not the best-case number in a footnote.

If you're trying to get past basin-level advisories and see whether a specific pad is emitting today, the daily, per-site plume mapping at Methane Plume Map is built around exactly that question. Request access to see what it resolves for a site you already have in mind.