Identifying spatial and temporal patterns of vegetation resilience across southern Africa using satellite time series data

摘要

Future changes in vegetation cover as a consequence of predicted climate change, or through anthropic stresses (e.g. fires, overgrazing, land abandonment) may have a substantial affect upon the world's ecosystems and important repercussions for the global carbon, energy and hydrological cycles. Understanding how environmental change may affect the world's arid and semi-arid regions is of particular interest given that such environments occupy as much as 40% of the global land surface and provide food resources for one sixth of the world's population. Southern Africa, like much of the African content, is considered highly vulnerable to future projected climatic change. In this region such changes are occurring alongside a number of developmental stresses, such as food insecurity, poverty, and the land tenure legacy of apartheid. Understanding these factors, along with potential stresses associated with climate change is increasingly recognised as integral to policy making and climate change adaptation. The characterisation of vegetation cover dynamics is thus required to better understand vegetation responses to human or climate induced change and to aid in the management of sensitive ecosystems. Vegetation "persistence" is an important concept for understanding land cover dynamics as knowledge of persistence characteristics can provide an indication of vegetation stability and thus the likely resilience of an ecosystem to change. Persistence analysis is able to identify both long-term stationary and non-stationary vegetation dynamics, the former being indicative of vegetation that is able to recover from environmental perturbations and maintain a stable regime, whilst the latter may be linked to climatic or anthropic stress. In an attempt to further our understanding of vegetation dynamics across the Southern African continent, we used AVHRR-NDVI bi-monthly time series data, derived from the GIMSS (Global Inventory Modeling and Mapping Studies) dataset (8 km resolution) to investigate vegetation persistence and stability across the Southern African countries of Botswana, South Africa and Lesotho, for a 24 year period (1982-2006). We estimated the persistence probability of positive and negative trends across the African countries and specifically how land cover and topography influence the spatial pattern of vegetation stability. We found that spatial patterns of both positive and negative trends showed clear evidence of spatial coherence, indicating collective dynamic behaviours of vegetation cover. Large clusters of positive trends were observed within the central parts of South Africa and Botswana. Conversely, clusters of negative trends were most apparent across central and western portions of Namibia, eastern and southern South Africa and Swaziland. Generally, the mean recovery times from negative trends for were shorter than those estimated for positive trends, which is as expected for healthy vegetation. However, evidence of possible weak resilience was observed at higher altitudes and within several key biomes (e.g. Succulent Karoo). The results suggest that AVHRR-NDVI time series data can be used for capturing details of vegetation cover dynamics and, when linked appropriately with climatic and anthropic drivers of environmental disturbance, can provide insights into the complex interactions between land cover, environment and climate, potentially aiding remedial approaches.