How tidal stage changes the mangrove canopy boundary in satellite imagery

Pull two satellite scenes of the same mangrove fringe, one at low tide and one near high water, and you'll get two different edges. Same trees, same year, different polygon. If you've ever had a reviewer ask why this year's extent map doesn't quite match last year's at the seaward fringe, tidal stage is usually the answer before canopy loss ever enters the picture.

Why the edge moves with the water

Mangrove canopy itself doesn't move with the tide, but what a sensor can see of it does. At low tide, exposed prop roots, pneumatophores, and bare mudflat sit in full view at the seaward margin, and a classifier can pick up root structure or wet sediment as something other than clean canopy. At high tide, water can push up into gaps between trees and along tidal creeks, and in a sparse or recently disturbed fringe, that water surface can read as open water rather than canopy, even where there's live mangrove above it.

The result is a boundary that's partly a function of canopy and partly a function of where the waterline happened to sit when the sensor passed overhead. For a dense, mature stand this washes out. For a thin fringe, a dieback zone, or a regrowth area with low, gappy canopy, it can shift the mapped edge by a noticeable margin, enough to read as loss or gain if nobody's checking the tide.

Local hydrology compounds this. A scene timed to a spring tide behaves differently than one timed to a neap tide even at the same nominal "low water" label, and a site with a flashy tidal range will show more boundary movement between stages than a site on a muted, mesotidal coast. This is also where sediment plumes and turbidity come in: a falling tide near a river mouth kicks up sediment that can change water's spectral signature enough to confuse a water/vegetation threshold near the edge.

Why tidal datum matters for a registry report

A registry reviewer looking at a multi-year extent time series isn't just checking whether there's still mangrove here. They're checking whether the boundary is tracking real change or tracking which tide happened to be in when each year's scene was captured. If year one was mapped near low water and year three was mapped mid-tide, a stable fringe can show up as loss on paper, and that's the kind of discrepancy that turns a routine submission into a round of clarifying questions.

Pick a tidal stage, tie it to a datum where you can (mean low water, mean sea level, whatever your project area's tide station reports), and match that stage scene to scene, year over year. Noting the tide state and, where available, the predicted height at acquisition time in the methods section of a monitoring report gives a reviewer something concrete to check the boundary against, rather than an assumption to take on faith. It also means that if a reviewer does flag a boundary shift, you can point to a tide mismatch as the explanation instead of scrambling to re-derive it after the fact.

This is also why cloud-cover-driven scene selection is such a headache for extent mapping specifically. A team that grabs whatever cloud-free pass is available that quarter has no control over tidal stage at all, and over several years of reporting cycles that produces a boundary that wanders for reasons that have nothing to do with the mangrove. Locking the acquisition to both a cloud-free window and a target tidal stage is what keeps a multi-year extent series comparable.

Mangrove Monitor builds its annual extent layer around exactly this kind of matched acquisition, so canopy boundary, loss, and regrowth polygons for the current reporting year sit on the same footing as the ones from last year's report. If your project's extent map has been drifting for reasons you can't quite explain to a reviewer, it's worth checking the tide before you chase a loss signal that might not be there.