Estimating fill volume from before-and-after satellite imagery

A concession report lands with a cubic meter figure for reclaimed fill, three months after the last one, and the EPC contractor's progress claim doesn't match it. The instinct is to pull two satellite passes, eyeball the new ground, and back into a volume. Two flat images will give you footprint change reliably: how much new ground appeared between the two dates. They won't give you volume on their own, because volume needs a height surface across the fill area, and a single optical pass doesn't carry elevation.

Area is not volume

Two image dates, before and after, give you a footprint: the reclaimed polygon grew by some number of hectares, the spoil ground pushed out another fifty meters past the old toe line. That's a real, measurable change, and it's the thing optical satellite imagery is good at.

Volume is a different question. It needs a height, or more precisely a height surface across the whole fill area, because reclamation lifts aren't flat and spoil mounds aren't uniform. Comparing two flat images side by side shows the footprint change; getting thickness at any given point requires an elevation surface the images themselves don't carry. Treating footprint growth as a volume number is the mistake that gets an estimate rejected by whoever's checking the math on the other end, usually the harbor engineer or the dredging contractor's QS.

The two ways to get an actual volume

If the deliverable has to be cubic meters or tonnage, there are really only two routes that hold up.

The first is a digital elevation model built from stereo or tri-stereo satellite pairs, or from a proper topographic survey, at both dates. Subtract the earlier surface from the later one across the fill footprint and you get a volume raster, cell by cell. This is standard practice for stockpile and borrow pit auditing in mining, and ports use the same math for reclamation fill and spoil disposal mounds. It needs stereo capability or a survey-grade DEM, not a single optical pass.

The second is cruder and depends on what you already know about the fill. If the design lift height is fixed, say a one-meter compacted layer per phase per the reclamation design drawings, then area added times known lift height gives a defensible volume for that phase. This only works when the engineering is settled ground: you need the as-built lift thickness from the contractor's method statement or the concession's own survey, not a guess. Multiply, apply a bulking or compaction factor if you're converting to loose tonnage, and you have a number worth putting in a report, caveated as design-based rather than measured.

Bathymetric or hydrographic survey data for the dredge source, where spoil volume is being tracked at the disposal ground rather than the reclamation side, is its own separate measurement and doesn't come from optical imagery at all.

What two passes of RGB imagery do reliably

Comparing a new quarterly pass against the last one is a solid way to catch footprint change: reclaimed area added since the last quarter, berth line extended by a given run length, spoil ground growing out past its prior boundary. That footprint change is the leading indicator worth tracking in its own right, apart from whatever volume figure eventually gets attached to it, and it's the thing that otherwise only shows up in an operator's press release or an infrequent site visit.

Port Monitoring is built around exactly that comparison: each new quarterly pass checked against the last, with the reclaimed area, berth extension, and spoil ground growth delivered as a change layer with before/after image pairs, exportable as GeoJSON. It flags where the fill footprint moved. Pairing that change layer with a known lift height or an occasional survey-grade elevation check is how you turn footprint growth into a volume figure you can defend, rather than skipping straight from two pictures to a tonnage number nobody asked you to justify.

If what you need quarter to quarter is confirmation that the reclamation footprint or the spoil ground actually grew, and by how much area, that's the gap this is built to close.