Abstract: Analyzing vegetation phenology is the key step for the accuracy evaluation of extraction methods using remote sensing. There is no consensus about the advantages and disadvantages of different methods. It is necessary to comprehensively evaluate the accuracy of various extraction methods. In this study, 18 combining methods for vegetation phenology remote sensing extraction and observation data from 23 ground phenology stations were used to evaluate the accuracy in extracting three key vegetation phenophases: Start Of the growing Season (SOS), End Of the growing Season (EOS) and Length Of the growing Season (LOS). The results showed that SG_{Sa 0.1} (the combining Savitzky-Golay filter and Seasonal Amplitude method with a threshold of 0.1) had the optimal recognition effect. For single phenophase, the optimal combination for SOS, EOS and LOS was SG_{Sa 0.1}, SG_{Sa 0.3} and DL_{Sa 0.1}, respectively. The minimum deviation between the optimal extraction results and the groundbased observation phenology data were more than five days, indicating the accuracy of extraction phenology using remote sensing data with 8-day temporal resolution. The deviation highlighted the importance of choosing a suitable extraction combination. In addition, remote sensing phenology and groundbased phenology cannot completely match. Therefore, the multiple evaluation indices used in this study, especially the deviation consistency can effectively select the optimal vegetation phenology remote sensing extraction combination for various phenophases.

Key words: sulfur, heavy metal, soil, phytodetoxification, speciation transformation.