**Abstract:**There is a significant negative correlation between RUE and rainfall and RUE fluctuates wildly from year to year so that RUE, used in isolation, says as much about rainfall variability as about land degradation. To avoid the correlations between RUE and rainfall, and to distinguish land degradation from the effects of rainfall variability, Wessels and others (2007) suggest the alternative use of Residual Trends (RESTREND). Following their general procedure, we have correlated for each pixel annual sum NDVI and annual rainfall. The resulting regression equation represents the statistical association between observed sum NDVI and rainfall; the model predicts sum NDVI according to rainfall. Residuals of sum NDVI (i.e. differences between the observed and predicted sum NDVI) for each pixel were calculated, and the trend of these residuals (RESTREND) was analysed by linear regression.
RESTREND points in the same direction as RUE: a negative RESTREND may indicate land degradation, a positive RESTREND improvement but the spatial distribution is different from RUE; overall, RESTREND patterns are remarkably close to sum NDVI of lesser amplitude.