Abstract

Vegetation index (VI) derived from satellite data is a key indicator for researches on crop type identification, growth monitoring and yield forecasting. However, due to the influences of the disturbance factors including the cloud contamination, atmospheric variability and bi-directional effects, the time series VI data are always with significant fluctuations. Such limitations greatly constrain the further application of time-series VI data. How to reconstruct high-quality time-series VI data has become a challenge especially for regional study. In this study, an operational method based on Savitzky–Golay filter was adopted for reconstructing MODIS Enhanced Vegetation Index (EVI) data from 2000 to 2008 in the Haihe River Basin in North China. The reconstructed EVI time-series dataset was analyzed together with land-use map and evapotranspiration data. It is found that this method can significantly reduce the containments of clouds and some other abnormal noises, and can greatly improve the quality of EVI time-series dataset both in space and time. It is also noted that this approach shows limitations for some small built-up lands dotted in vegetation land. We discuss the possible reasons in further research.

Key words: Vegetation index, crop monitoring, Savitzky–Golay filter.