Land degradation threatens ecosystems and socio-economic development. Degradation detection is imperative for understanding and preventing degradation. Three indicators have usually been used to detect land degradation: NDVI trends for visible degradation, residual for human-induced potential degradation, and sensitivity of vegetation to rainfall for functional degradation. However, we remain unclear about the relationship between these indicators. In this study, we applied methods of linear regression, Time Series Segmented and Residual Trend (TSS-RESTREND), and Sequential Regression (SeRGs) to detect degradation in southern Africa and compared them. The data was derived from GIMMS 3.1 NDVI dataset and MSWEP rainfall dataset. The results showed different degradation proportions ranging from 7.1% to 18.7%. The co-occurrence of degradation only occupied 3.84%, whereas single degradation occupied 27.21% of the total area. The potential and functional degradation still existed in the regions with vegetation greening. These results indicated the discrepancies and divergent applicability among these indicators. Compared with previous cases, the relationships between indicators were clarified, in terms of drivers, processes of degradation, and state of the ecosystem. This enhanced the understanding of multifaceted aspects of degradation. On the regional scale, the degradation is significantly divergent across aridity (e.g. as aridity increases the proportions of functional degradation increase) and countries (e.g. degradation combination B in BWN’s forest, shrubland, and grassland is higher than others). These provided the assessment of the effect of policies on degradation and the reference to prevent degradation under a certainty arid zone. Generally, this study indicated the necessity of degradation detection with multiple methods.
 

Degradation Detection, TSS-RESTREND, SeRGs, Dryland, Southern Africa