Abstract:

There is a crucial need in global change studies to understand how terrestrial ecosystems respond to climate systems. It has been demonstrated by many researchers that the Normalized Different Vegetation Index (NDVI) time series from remotely sensed data, which provide effective information of vegetation conditions at large scales with high temporal resolution, are closely correlated with meteorological factors. However, few of these studies have taken the cumulative property of NDVI time series into account. In this study, NDVI difference series was proposed to replace the original NDVI time series to reappraise the relationship between NDVI and meteorological factors. As a proxy of vegetation growing processes, NDVI difference represented net primary productivity (NPP) of vegetation during specific time intervals and under specific environmental conditions. This data replacement eliminates cumulative effects that exist in the original NDVI time series, and thus is more appropriate for understanding how climate systems affect vegetation growth over short time scales. Using correlation analysis, we studied the relationship between NOAA/AVHRR ten-day NDVI difference series and corresponding meteorological data from 1983 to 1999 from 11 meteorological stations located in the Xilingole steppe in Inner Mongolia. Our analyses showed the following results. 1) Meteorological factors were found to be more strongly correlated with NDVI difference at the biomass-increasing phase than during the decreasing phase. 2) The relationship between NDVI difference and climate variables varied with vegetation types. For a typical steppe community dominated by Leymus chinensis, temperature had a higher correlation with NDVI difference than precipitation and, for a typical steppe community dominated by Stipa krylovii, the correlation between temperature and NDVI difference was lower than that for precipitation. For a typical steppe dominated by S. grandis, there were no significant differences between the two correlations. Precipitation was the key factor influencing vegetation growth in desert steppe communities, and temperature had a poor correlation with NDVI difference. 3) The response of NDVI difference to precipitation is fast and almost simultaneous in both typical steppe and desert steppe; however, mean temperature exhibited a time-lag effect, especially in the desert steppe and some typical steppe ecosystems dominated by S. krylovii. 4) The relationship between NDVI difference and temperature is becoming stronger with global warming.

Key words: NDVI difference time series, Autocorrelation, Cumulation, Correlation analysis, Grassland, Precipitation, Temperature