How to tell a looting pit from a utility trench in satellite imagery

Every monitoring officer who has scrolled through a quarter's worth of VHR tiles has hit the same moment of doubt: a dark patch of disturbed ground near the tell, and no way to say on sight whether it's a looter's pit or a contractor's trench for a water line. Both show up as fresh soil disturbance against an otherwise stable background. Both can appear overnight between archive passes. The difference is in the geometry, not the color.
Shape and spacing tell you more than darkness does
A looting pit is almost never a straight line. Looters dig where they expect a burial chamber, a foundation course, or a cache, so the pits cluster in irregular groups that follow the archaeology, not the terrain. Diameters commonly run a meter or two across, sometimes larger on a known tomb mound, and the spacing between pits in a cluster is uneven: tight in one corner, scattered a few meters off in another.
A utility trench runs differently. It's linear, holds a near-constant width along its length, and follows a right-of-way: a road edge, a property boundary, the shortest path between two points of connection. A disturbance that traces a straight or gently curved line for fifty meters and stops at a junction box or a building wall is infrastructure work.
Spoil placement and backfill timing
Spoil is the giveaway most officers learn to check second. Looting spoil tends to sit in an irregular halo around the pit mouth, thrown rather than placed, because there's no equipment and no reason to stack it neatly. It often stays visible for more than one imaging pass since nobody is paid to clean it up.
Trench spoil is usually a tidy windrow running parallel to the cut, sometimes with visible equipment tracks beside it from a backhoe or a roller. And it disappears fast. Utility work gets backfilled and graded within days to weeks, so if you see the same linear scar still open two or three revisit cycles later, reconsider the utility explanation; abandoned infrastructure projects happen, but so does looting that nobody has flagged yet.
Context clues worth checking before you log a disturbance
Location relative to known site boundaries matters more than either shape or spoil on its own. A cluster of irregular pits inside a scheduled monument boundary, away from any road or existing utility corridor, is looting until proven otherwise. A linear cut that aligns with a visible road, a power line easement, or a prior utility corridor you've seen in earlier imagery is almost certainly maintenance work, even if it happens to run near a protected zone.
Vehicle tracks are another tell, and they cut both ways. A looting operation working at scale leaves tracks that approach the pit cluster and then stop, with no onward connection to anything. A utility crew leaves tracks that connect to a road, run along the trench line, and often show a parked-vehicle footprint where equipment sat for several days.
If none of this resolves the call from a single frame, don't force it. Pull the prior quarter's tile and the current one side by side. A disturbance that wasn't there last cycle and has the pit cluster signature, not the trench signature, is the one worth writing up for the quarterly report.
When the two still look alike
Some disturbances sit right on the line: a narrow irregular cut near a site boundary that could be a small looting trench cut by hand, or could be a utility repair that never got properly backfilled. In those cases the call comes down to pattern over time rather than a single image. One pit or one short cut might be ambiguous. A site that keeps producing the same pit-cluster signature quarter after quarter, regardless of what utility work is also happening nearby, is the pattern a monitoring report should carry forward.
That kind of quarter-over-quarter read is exactly what repeated manual comparison is bad at and what Looting Monitoring was built to do instead: a dated before/after pair for the same ground, with the new disturbance already flagged so the shape-and-spoil judgment call is yours, not buried in a raw image dump.
If your next review cycle has a site with disturbance you can't yet classify, that's the comparison worth running first.