Monitoring vegetation phenology in China using time-series MODIS LAI data

Land surface phenology dynamics reflect the response of terrestrial ecosystems to inter- and intra-annual dynamics of the climate. However, there are very few regional-to-global phenology products and existing phenology products still show some deficiencies in practical application. Meanwhile, none of existing methods for monitoring vegetation phenology has consistent performance for all vegetation types. Based on the existed research work, this paper developed a mixed model to monitor vegetation phenology in China. Different methods are used in this model according to different situation. This model was employed in China in 2007 assessed using field observed phenology and MLCD data. The root mean square error (RMSE) for different vegetation types are 4.0-33.5, the mean absolute error are -20.6-15.3 and the correlation are 0.404-0.887. By comparison with MLCD data, the success rate and the accuracy of the method have been highly improved.