Habitat Loss Mapping

Blog

What is land degradation neutrality under SDG indicator 15.3.1?

When a country reports under SDG indicator 15.3.1, it isn't reporting one number. It's reporting a status, degraded, improved, or stable, built from three separate sub-indicators that each measure something different about the land. Land degradation neutrality (LDN) is the target itself: a state where the stock of healthy, productive land doesn't shrink over a reporting period, even as individual parcels shift from one use to another. Indicator 15.3.1 is the mechanism the UNCCD and the SDG framework use to ask countries to show they're tracking toward it.

The three sub-indicators, and why land cover carries the most weight

The proportion of degraded land is assessed against three sub-indicators: land cover and land cover change, land productivity dynamics, and carbon stocks, usually estimated through soil organic carbon. A pixel counts as degraded if it shows negative change on any one of the three. That's the "one-out, all-out" rule the UNCCD lays out in its Good Practice Guidance.

Land productivity dynamics and soil organic carbon are typically modeled from global datasets, Trends.Earth, NDVI trend products, SoilGrids, that most national statistics offices don't generate in-house. Land cover is different. It's the sub-indicator you can build, inspect, and defend against your own national baseline, because it's a direct classification of what's physically on the ground: forest converted to cropland, cropland converted to built-up, wetland lost to agriculture. When a reviewer asks why a given hectare flipped from stable to degraded, the land cover transition matrix is the part of the answer you can trace back to a dated image.

Where land cover reporting breaks down between cycles

Most of the trouble with the land cover sub-indicator isn't the classification scheme, FAO's LCCS or whatever national legend you've crosswalked to it. It's consistency between reporting cycles. The UNCCD cycle runs roughly every four years, and countries compare the current period against a fixed baseline, commonly 2015, plus the prior reporting period. If your land cover layer for cycle two was built with different imagery, a different classifier, or a different contractor than cycle one, the "change" you report is partly real and partly noise from methodology drift. A reviewer asking why the forest-to-cropland transition rate doubled between cycles, with nothing on the ground changing that dramatically, has found a problem with your baseline, not with the land.

Sub-national reporting adds another layer. Many ministries now produce LDN figures by province or district to support domestic land-use planning alongside the national SDG return, which means the land cover layer needs to hold up at a finer administrative scale without a separate commissioning exercise for every region, every cycle.

What holds up at the next reporting cycle

For the land cover sub-indicator specifically, what the reporting file needs most is the same classification logic run at the same cadence, far enough back to establish a real baseline, at a resolution that resolves the classes your legend distinguishes. None of that requires a bespoke accuracy claim for your territory. It requires a layer you don't have to re-explain to the methodology reviewer every four years because the sensor, the vendor, or the classification approach changed underneath you.

That continuity is the gap most statistics teams hit when the original land cover study was a one-off consultancy deliverable: it answered cycle one, and cycle two needs its own tender, its own classifier, its own set of assumptions written up from scratch. Habitat Loss Mapping was built for that specific gap, an annually-refreshed land cover change layer over a national or sub-national territory from the same wide-swath multispectral source, so the transition matrix you hand to the UNCCD reporting portal this cycle was produced the same way as the one you handed over last cycle.

If your land cover sub-indicator is still rebuilt from scratch every reporting period, it's worth working out what a standing, year-over-year layer would change about that conversation.

Get started

← Back to the blog