Satellite-Derived Productivity Outputs for Land Degradation Assessment Vary With Biome and Rainfall

Abstract

Estimates of the extent of land degradation vary dramatically. To quantify the extent and location of land degradation globally, the United Nations Convention to Combat Desertification (UNCCD) asks member countries to report on land degradation every four years, providing crucial baseline and change data. Finding remotely sensed products that best capture degradation is central to this reporting. Here, we compare assessments for South Africa's Trajectory of Land Productivity using the Trends.Earth tool, which allows use of various remotely sensed data and climate inputs (Land Productivity Degradation Models, LPDMs). These differ in how they account for precipitation, allowing countries to choose the most appropriate for their context. We compare extent and location of degraded, stable or improved Productivity indicator pixels as identified by five different LPDMs for the country over the UNCCD 2022 reporting period (2016–2019), and whether this varied with biome. The LPDMs differed in percentage of area identified as degraded. The most pessimistic identified 35% degraded and 4% improved, the most optimistic, 15% as degraded, and > 70% improved. LPDMs also differed in where degradation was identified. Models that account for rainfall were more likely to classify a location favorably than those that did not account for rainfall, particularly in shrubland biomes. South Africa's Grasslands were less likely than other biomes to be classified as degraded. Between 59%–78% of the country's area experienced drought over the reporting period, which may have accentuated differences between LPDM outputs. Mapping and monitoring degradation over space and time is crucial; to achieve this, all LPDM outputs should be carefully assessed by ecologists with a working knowledge of the landscapes of interest, supported by field validation data. This approach ensures that the most suitable remotely sensed models are used in national monitoring and reporting.