Spatial Response of Vegetation to Precipitation in Dry Lands of Kazakhstan: Combination of Remote Sensing Data with Climate Records

Abstract

This study analysed spatial responses of vegetation to precipitation using 10-day images of Normalized Difference Vegetation Index (NDVI) derived from the Advanced Very High Resolution Radiometer (AVHRR) and rainfall records collected at 9 climate stations throughout a large semi-arid region in Central Kazakhstan. The response of NDVI to precipitation was estimated by calculating coeffi cients of spatial correlation between corresponding gridded maps of NDVI and rainfall during the period 1985-2001.The analyses were carried out at two different spatial scales (the whole study region and the scale of individual vegetation type) and two temporal scales (annual and within each of the individual growing seasons). The results proofed a strong relationship between
NDVI and precipitation: NDVI and precipitation co-varied in the same direction (either positive or negative) at all spatial and temporal scales upon study. However, the response of vegetation to precipitation exhibited signifi cant spatial and temporal variability.At the scale of the whole growing season, the NDVI-rainfall correlation increases from desert to semi-desert and steppe vegetation.On the contrary, at the within-season scale, desert vegetation demonstrated a much stronger dependence on precipitation than other two vegetation types. The relationships between NDVI and precipitation were found to be strongly non-linear: the upper threshold for the linear relationship is about 250 mm for growing season and 15-30 mm for 10-day rainfall amount. Above this threshold, the
response of vegetation to precipitation substantively decreases.