Cheatgrass vs. native bunchgrass: telling them apart in satellite imagery
If you manage range or wildland units in the Intermountain West, you already know cheatgrass by eye once it cures out red-purple in early summer, weeks ahead of the native perennials still holding green. The question is whether that same contrast shows up reliably enough in imagery to map it, instead of sending a crew to walk transects and eyeball every drainage.
It does, but only if you catch the right window and have enough spatial and spectral detail to separate a cheatgrass patch from a bunchgrass clump sitting right next to it.
Why timing beats color alone
Cheatgrass (Bromus tectorum) is a winter annual. It germinates in fall, greens up fast in spring, flowers and sets seed early, then cures to that reddish-tan color well before native bunchgrasses like bluebunch wheatgrass or Idaho fescue finish their growing season. On the ground this is the "red stage" range techs have used for decades as a rough field cue.
In imagery, the useful signal isn't the color itself, it's the mismatch in timing. During the cheatgrass green-up window, a stand of cheatgrass and a stand of native bunchgrass can look almost identical in a single RGB image, both green, both grassy, same rough texture. The separation shows up when you compare dates: cheatgrass senesces and loses chlorophyll on its own early schedule while the natives are still photosynthesizing. Catch a scene in that gap and the reflectance difference, particularly in the red and near-infrared bands, becomes much easier to pull apart than it would be from a single midsummer flyover.
This is why a single cloud-free satellite pass from any old archive rarely settles the question. You need an image timed to the phenology gap, not just a clear day.
What the spectral signature actually looks like
Cured cheatgrass has lower chlorophyll absorption in the red band and a different near-infrared response than still-green native grasses, because one canopy is dying back and the other isn't. That shift shows up as a change in the red-edge position, the narrow band where reflectance climbs steeply from red into near-infrared as a canopy senesces. A hyperspectral sensor resolves that edge with enough bands to flag it as a distinct curve shape, not just "redder" or "duller" than its neighbor. A standard four-band RGB-NIR satellite image mostly can't make that distinction with any confidence, which is the gap that trips up a lot of DIY mapping attempts built on whatever free imagery happens to be available.
Resolution matters just as much as spectral detail. Cheatgrass rarely shows up as a clean, solid-color block. It infiltrates between bunchgrass clumps, fills interspaces after a burn, and threads along a two-track or a fence line before spreading into the unit. At 10 m or 30 m pixels, that patch gets averaged in with everything around it and disappears into "grassland, mixed." You need sub-meter imagery to hold the edge of a patch against the bunchgrass it's displacing. That's the difference between a map a crew can walk to and a county-scale heat map that tells you cheatgrass exists somewhere in the watershed, which you probably already knew.
Putting it together for crew work
None of this replaces a range tech's eye in the field. What it does is narrow where that eye needs to go first. A unit might run a few thousand acres, and a visual sweep or a walking transect survey takes days, with a real chance of missing scattered or low-density patches that haven't dominated yet. A map built from VHR hyperspectral imagery captured at the right phenology window gives you patch boundaries before the crew ever drives out, so treatment goes to the infestation instead of the whole allotment.
Invasive Mapping builds that seasonal map for a species you name, timed to the window where it separates from the natives around it, so your crews start the season with coordinates instead of a windshield survey.
If cheatgrass is the problem species on your unit this year, it's worth seeing what that map would show for your ground.