Methodology
From visible greening to functional regeneration.
The LMDC framework compares multiple forms of evidence to distinguish surface change from deeper ecological, hydrological and climatic trajectories. It is a first iteration — a tool to be tested, criticised and improved, not a closed standard.
Explore the levels
Four levels, increasing evidence requirements
Select a level to see its purpose, indicators and limits. Progressing from N1 to N4 is not automatic — most projects plateau where the documentation runs out and measurement would need to begin.
Purpose
Establishes the existence of the project and its parameters. Documentary, and generally straightforward to complete.
Indicators
Hydrography & topography, pedology, rainfall regime, cost per hectare, governance & local anchoring, economic model.
Typical data sources
Terrain models, station/satellite rainfall records, programme documentation, published reports.
Confidence ceiling
High for existence and parameters; some indicators (e.g. cost per hectare) still require harmonisation across sources.
What it can demonstrate
That a project exists, where, since when, under what governance and at what declared cost.
What it cannot demonstrate
Whether any ecological, hydrological or climatic function has actually changed.
Example
Every project in the directory reaches N1 — it is the baseline for inclusion. See Mu Us Sandy Land's N1 row.
Purpose
Assesses vegetation performance and ecological function — the level most people mean by "greening".
Indicators
Vegetation performance over the longest series the site allows; other ecological results where the literature reports them, after normalisation.
Typical data sources
Satellite vegetation indices, field ecological surveys, published time series.
Confidence ceiling
Moderate to high once a produced (not merely compiled) measurement exists; metrics vary study to study and need normalising.
What it can demonstrate
Whether cover is increasing, and since when, at project scale.
What it cannot demonstrate
Whether the greening reflects a real function returning or a Goodhart's-law artefact — an intensive practice can raise a vegetation index without enriching the system.
Example
At Niakhar, tree cover is measured as a real fertility island (about 23% of primary production), not inferred from greenness alone. See the project.
Purpose
The pivot level: groundwater stocks, water flows and surface energy exchange. This is where documentation alone almost always runs out.
Indicators
Groundwater level / stock, surface energy flux assessment (evaporative cooling vs. radiative heating), dew and non-rainfall moisture.
Typical data sources
Instrumented monitoring (piezometers, flux towers), satellite gravimetry, targeted field campaigns — rarely literature alone.
Confidence ceiling
Low by default until measurement is produced; most projects show empty cells here, which the grid treats as the documentary ceiling, not a lapse.
What it can demonstrate
Whether water and energy balances are genuinely shifting, and whether apparent gains are a flow (renewable) or a drawdown of stock.
What it cannot demonstrate
Regional or climatic attribution on its own — that requires N4.
Example
At Niakhar, a flux tower has instrumented surface energy exchange in situ since 2018 — one of the few sites in this directory with a produced, not merely compiled, N3 measurement.
Purpose
Places the project within the region's climatic trend, and — as a separate, much harder question — asks whether the project itself measurably affects that trend.
Indicators
Regional climate drift, read from the literature regardless of project size.
Typical data sources
Climate reanalyses, regional attribution studies, long-run station networks.
Confidence ceiling
The regional trend itself is frequently contested between methods (see the Loess Plateau and Sahel cases in the methodology report); project-level attribution is a research horizon, not a delivered indicator.
What it can demonstrate
Where a project sits inside a documented regional trend.
What it cannot demonstrate
By itself, that the project caused any part of that trend — see the two meanings of N4 below.
Example
Mu Us sits within a regional rainfall trend that is itself disputed between reanalysis products — the project is placed in that debate, not used to resolve it.
A necessary distinction
N4 contains two different questions
Conflating them is the single most common way a regeneration claim overreaches. LMDC keeps them visually and structurally separate.
In contract
The regional climate trend around the project
A literature-based reading of how the region's climate is already changing, independent of the project. This contextualises the project inside a broader trend without crediting or blaming the project for it. It can be read from bibliography alone, regardless of project size.
Out of contract — research horizon
Whether the project itself affects the climate
The measured effect of the intervention on local or regional climate, and any feedback loops. This is a research hypothesis requiring substantial area and instrumentation. It appears in case narratives (e.g. the three-phase Mu Us trajectory) — never as a filled-in indicator cell.
LMDC does not provide automatic causal proof at N4. It organises what the literature already establishes, isolates what remains to be measured, and makes the boundary between the two visible.
Full methodology
Operational scope and architecture
Dryland eligibility & the aridity index
Deserts.eco follows the UNCCD's broad definition: drylands are territories where the aridity index P/PET remains below 0.65. Hyper-arid land (below 0.05) is excluded. The index is read on a rolling, multi-year basis — a site near the 0.65 boundary is not permanently in or out based on a single wet or dry year.
Documented ecosystem thresholds fall near aridity (1 − P/PET) values of 0.54, 0.70 and 0.80 — i.e. P/PET ≈ 0.46, 0.30 and 0.20 (Berdugo et al. 2020).
The wetland-in-dryland exception
Some sites are arid by the index but fed by water sourced elsewhere — the limiting factor becomes upstream water management rather than local rainfall. Cebolla Canyon (a spring-fed canyon system) is read this way in this directory.
The attribution problem
A location is never observed both with and without an intervention — attribution rests on a counterfactual that must be constructed, not observed directly. Climate variability, concurrent forcings and divergent measurement methods (e.g. reanalysis versus rain-gauge networks) can all move the same signal in different directions. Global-scale studies can separate land-use change from broader climate change; local, single-project attribution is far harder and generally exceeds what a first documentary pass can establish.
Trajectory types
- Aggradation — a variable, or several, moving in a favourable direction.
- Maintenance / stabilisation — functions holding steady, sometimes against a degrading regional backdrop.
- Reversal — a documented decline followed by recovery, or vice versa (see Mu Us).
- Decoupling — surface cover recovers while underlying function does not (see Bowen Basin).
A negative trajectory or a reversal is a result, not a failure to hide — the literature under-documents failed projects, and LMDC treats that gap as a known bias rather than papering over it.
Spatial and temporal scale
N1 and N2 read at project scale. N3 requires instrumentation, often at a sub-project or plot scale for practicality. N4's regional-trend half reads at whatever scale the surrounding literature covers; its project-attribution half has no fixed area threshold in the research literature — effect depends on the spatial structure of the surface contrast (extent and configuration together), not extent alone.
Limitations
This is a first iteration. It compares many projects on common indicators — a reading that is strong for situating and comparing projects, but does not by itself resolve fine-grained, site-specific attribution, which requires dedicated field measurement. The framework also inherits a degraded historical baseline: much of the 20th-century agronomic shock to soils predates the satellite era, so reference states are likely themselves degraded — this directory measures forward trajectories rather than a return to an assumed prior state.
Research horizons
Several questions are acknowledged as open rather than answered: whether many small vegetation patches outperform one large one at landscape scale (established at plot scale, not yet at landscape scale); the cascade linking plot-level water balance to regional hydrology; and atmospheric moisture recycling, for which instrumentation is still described as lacking even by specialist bodies (Norman et al. 2026, USGS).
Volet A — Methodological framework
An English HTML overview is available on the resource page; the source PDF is French.