Comparative study on phenology in a mountainous plantation in northern China based on vegetation index, chlorophyll fluorescence and carbon flux

  Objective  Phenology is the rhythmic variation of vegetation development. It is a long adapting result to climatic and environmental changes. In this study, we investigated the relationships between MODIS NDVI, EVI, GOME-2 SIF and gross primary productivity (GPP) in order to explore the ability of reflecting dynamic variation of the vegetation. This study can provide some suggestions for deep understanding the reaction of the plantation to climatic changes.

  Method  The time series of NDVI, EVI, SIF and GPP from 2007 to 2011 were filtered using Timasat 3.3 software. Four types of index curves were fitted using double logistic functions and phenological parameters were extracted based on the maximum rate of change. Linear regression and root mean squared error (RMSE) between NDVI, EVI, SIF and GPP were used to compare the ability of reflecting the dynamic variations of the vegetation.

  Result  (1) The time series of MODIS NDVI, EVI and GOME-2 SIF were consistent with GPP. Monthly mean NDVI, EVI and SIF were positively correlated with monthly mean GPP from 2007 to 2011. (2) The determination coefficient (R²) values between monthly mean GPP and NDVI, EVI in spring and autumn were better than that between monthly mean SIF and GPP. However, GPP was strongly correlated with NDVI in summer. The relationships between GPP and EVI, SIF were not obvious. (3) RMSE between GPP-based phenological parameters and EVI-based phenological parameters was less than that between GPP and NDVI, SIF-based phenological parameters. The order of the RMSE value was EVI, NDVI and SIF.

  Conclusion  NDVI and EVI can better reflect the dynamic variation of the plantation in this site. The start of the growing season (SOS) extracted by MODIS NDVI and EVI advanced in comparison with the GPP-based SOS. The end of growing season (EOS) extracted by MODIS NDVI and EVI delayed in comparison with the GPP-based EOS. The main reason was that phenological information based on MODIS NDVI and EVI was extracted according to spectral characteristics of canopy structure and leaf reflectance. The SOS and EOS extracted by SIF advanced compared with the GPP-based SOS and EOS. The footprint of GOME-2 SIF mismatched with the footprint of flux tower GPP, thus impacting the relationship between SIF and GPP.