Main content area

Reconstruction of a complete global time series of daily vegetation index trajectory from long-term AVHRR data

Zhang, Xiaoyang
Remote sensing of environment 2015 v.156 pp. 457-472
climate, data collection, growing season, logit analysis, normalized difference vegetation index, primary productivity, remote sensing, satellites, snowpack, time series analysis, vegetation cover
Normalized Difference Vegetation Index (NDVI) derived from Advanced Very High Resolution Radiometer (AVHRR) has been extensively used for examining long-term dynamics of the vegetated land surface and climate impacts because it provides the longest time series of global satellite measurements. However, the applications are significantly limited by the persistent presence of atmospheric contamination and snow cover in the NDVI time series. Several approaches have been developed for smoothing and fitting the time series of biweekly NDVI composites but the capabilities of depicting land surface dynamics are method dependent. Using a time series of daily EVI2 (two band enhanced vegetation index) from AVHRR long term data record (LTDR) (1982–1999), this study reconstructed a global dataset of spatially and temporally consistent and continuous daily vegetation index. Specifically, the EVI2 outliers were removed and missing observations were filled explicitly based on biophysical properties of vegetation growing cycles. The EVI2 temporal trajectory was then reconstructed using a hybrid piecewise logistic model which is biophysically meaningful in describing vegetation growth. Moreover, the confidence for each annual time series in each individual pixel was quantified by determining both the proportion of good quality satellite observations and the agreement index of model fitting during a vegetation growing season. Finally, verification was performed, which indicated that the reconstructed EVI2 trajectory reflects well the field observations of seasonal green vegetation cover and the flux tower measurements of interannual gross primary productivity variation.