Habitat Loss Mapping

Blog

What is SDG indicator 6.6.1, and how land-cover data backs a water-ecosystem report

Calm wetland shoreline with tall reeds and trees reflected in still water

SDG indicator 6.6.1 tracks change in the extent of water-related ecosystems over time. The basket it covers is wider than people expect on first read: lakes, rivers, aquifers, and wetlands, plus permanent and seasonal water bodies, vegetated wetland, and in the fuller methodology the river basins and mountain systems that feed them. UNEP is the custodian agency. The methodology asks countries to set extent for a baseline year and then track how that figure moves afterward.

For a statistics analyst sitting inside a ministry, the indicator itself isn't the hard part. The hard part is the data pipeline behind it: where "extent" comes from, how you defend a number that moves from one reporting cycle to the next, and what happens when the classification method used in year one isn't the method used in year five.

Why the land-cover layer is the real deliverable

UNEP's guidance for 6.6.1 leans on satellite-derived land-cover and surface water datasets because they're the only practical way to get consistent extent figures across a whole territory, repeated over time. A ground survey of wetland boundaries is slow and expensive to repeat every cycle, and it doesn't scale to a national territory without a large field budget. Remote sensing scales, provided the classification is done the same way each time.

That consistency requirement is where a lot of national reports run into trouble. If the land-cover classification used for the 2018 baseline came from one contractor with one set of class definitions, and the 2024 update came from a different study with different thresholds for what counts as wetland versus seasonally flooded cropland, the change figure reported is really measuring the gap between two methodologies, and a technical reviewer will find it.

The indicator's own documentation flags this: comparability across time depends on a stable classification scheme. For wetland extent reporting SDG purposes, that means the same spectral bands, the same minimum mapping unit, and the same decision rules for separating open water, vegetated wetland, and dry land, applied identically to this year's imagery as to last year's.

What the classification actually needs to carry

A land-cover layer built for 6.6.1 reporting needs to do a few specific things.

It needs multispectral imagery wide enough to cover the reporting territory in a manageable number of scenes, because stitching together dozens of narrow-swath tiles from different acquisition dates introduces its own inconsistency. It needs a cadence that matches the reporting cycle, usually annual, so you're not interpolating between a four-year-old classification and a six-year-old one. And it needs the open-water and vegetated-wetland classes held stable release to release, so a shrinking wetland shows up as a shrinking wetland rather than as noise from a reclassified pixel.

None of this replaces the national custodian process. UNEP's reporting template still asks for your own validation and your own narrative around drivers of change: drainage, conversion to agriculture, upstream abstraction, whatever applies locally. What the layer gives you is the extent figures to put in that template, cycle after cycle, without re-running a classification study from scratch each time and hoping it lines up with the last one.

Most ministries reuse whatever classification already exists, patch gaps where imagery was cloudy or missing, and hope the comparison still holds, rather than commissioning a fresh land-cover study every reporting year. An annually-refreshed land-cover change layer built on one consistent method gives you the figures the SDG reporting calendar actually needs: a baseline you can stand behind, and a new figure each cycle produced the same way as the last, so the change you report in water-related ecosystem extent reflects what happened on the ground, with the classification method held constant.

If the next reporting cycle is already on the calendar, it's worth a look at what Habitat Loss Mapping can do for the extent figures in it.

Get started

← Back to the blog