Publications

Lei, SG; Ren, LX; Bian, ZF (2016). Time-space characterization of vegetation in a semiarid mining area using empirical orthogonal function decomposition of MODIS NDVI time series. ENVIRONMENTAL EARTH SCIENCES, 75(6), 516.

Abstract
There are always different patterns of temporal variation determined by the heterogeneity of the surface coverage and influence factors at regional scale. Identification of the temporal patterns and positioning their spatial distribution are critical for rational use of nature resources and ecological protection. In this study, the empirical orthogonal function decomposition (EOFD) was proposed for the time-space characterization of vegetation in a semiarid mining area using 206 MODIS-NDVI images with a spatial resolution of 250 m and temporal resolution of 16 days. Four kinds of temporal patterns of the vegetation were decomposed, which are annual cycle, deserti-fication, increase and degradation. Moreover, the spatial distributions of the decomposed temporal patterns were mapped. As a result, a much stronger relationship was found between the NDVI and rainfall, which is the key factor controlling the vegetation growth in the study area. The spatial distribution of the impacts on the vegetation from the underground coal mining, open-pit mining, and vegetation reconstruction was identified with the EOFD method and verified by site investigations. These results demonstrated that the EOFD method can decompose the whole temporal variation of vegetation into several different patterns of variations and help to distinguish between the influences from natural factors or anthropogenic activities. The EOFD method is determined by the temporal variations of the study area without human intervention; therefore, it is a promising methodology for time-space characterization with the aid of satellite images time series at a large scale.

DOI:
10.1007/s12665-015-5122-z

ISSN:
1866-6280