RESUMO
The long-term series of geographic data and remote sensing data contain noise and perio-dic fluctuation. We used the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) to decompose the data of the normalized difference vegetation index (NDVI), precipitation, and temperature from 1982 to 2015 on per-pixels in the Loess Pla-teau to obtain residuals. Using the residual with less noise and periodic fluctuations, we examined the changes of NDVI and the relationship between NDVI and climatic factors. The results showed that the spatial change trend of NDVI was mainly increasing from 1982 to 2015 in the Loess Plateau. The significance of the change trend of residual NDVI (95.9%) was greater than the original NDVI (72.3%), with spatial variations. Temperature and precipitation could largely explain the changes in vegetation coverage. The proportions of areas with extremely significant positive and negative correlations between temperature and NDVI on the Loess Plateau were 83.7% and 13.9%, respectively, while that between precipitation and NDVI were 54.4% and 37.2%, respectively. There were obvious spatial variations in the responses of vegetation to climate change on the Loess Plateau. Different climatic factors had different effects on different types of vegetation. In general, temperature had stronger correlation with different vegetation than precipitation. Therefore, temperature was the main driving factor for the changes of vegetation cover in the Loess Plateau.
