SEARCH

SEARCH BY CITATION

Keywords:

  • Global land cover change;
  • MODIS;
  • multitemporal change vector method;
  • precipitation;
  • sub-Saharan Africa

ABSTRACT

Aim  Interannual land cover change plays a significant role in food security, ecosystem processes, and regional and global climate modelling. Measuring the magnitude and location and understanding the driving factors of interannual land cover change are therefore of utmost importance to improve our understanding and prediction of these impacts and to better differentiate between natural and human causes of land cover change. Despite advances in quantifying the magnitude of land cover change, the interpretation of the observed land cover change in terms of climatic, ecological and anthropogenic processes still remains a complex issue. In this paper, we map land cover change across sub-Saharan Africa and examine the influences of rainfall fluctuations on interannual change.

Location  The analysis was applied to sub-Saharan Africa.

Methods  Ten-day rainfall estimates (RFE) obtained from National Oceanic and Atmospheric Administration's (NOAA) Climate Prediction Center (CPC) were used to extract information on inter and intra-annual rainfall fluctuations. The magnitude of land cover change was quantified based on the multitemporal change vector method measuring year-to-year differences in bidirectional reflectance distribution function (BRDF) corrected 16-day enhanced vegetation index (EVI) data from the Moderate Resolution Imaging Spectro-radiometer (MODIS). Statistical models were used to estimate the relationship between short-term rainfall variability and the magnitude of land cover change. The analysis was stratified first by physiognomic vegetation type and second by chorological data on species distribution to gain insights into spatial variations in response to short-term rainfall fluctuations.

Results  The magnitude of land cover change was significantly related to rainfall variability at the 5% level. Stratification considerably strengthened the relationship between the magnitude of change and rainfall variability. Explanatory power of the models ranged from R2 = 0.22 for the unstratified model to 0.40–0.96 for the individual models stratified by patterns of species distribution. The total variability explained by the combined models including the influence of rainfall and differences in vegetation response ranged from 22% for the model not stratified by vegetation to 76% when stratified by chorological data.

Main conclusions  Using this methodology, we were able to measure the contribution of natural variation in precipitation to land cover change. Several ecosystems across sub-Saharan Africa are highly sensitive to short-term rainfall variability.