Dynamics of phenology and its response to climatic variables in a warm-temperate mixed plantation

Abstract

Vegetation phenology is an integrated indicator of vegetation growth and development in response to climate, and it is one of the important variables regulating carbon flux in forest ecosystems. Land surface phenology derived from different remote sensing products and their relationships with flux tower gross primary production (GPP) data have been paid much attention. Remotely sensed phenology was retrieved based on greenness, while GPP-based phenology was extracted based on photosynthesis. However, exploring the differences between remotely sensed and GPP-based phenology are limited. In this study, we compared phenological metrics obtained from flux tower GPP of a mixed plantation with phenological metrics derived using MODIS EVI and MCD12Q2 from 2006 to 2017. We also explored phenological transitions and the response of phenology to climatic variables. The results showed that the 16-day composited EVI time series was in agreement with the 16-day composited GPP time series (R2 = 0.86, p < 0.001). The root mean squared deviation values between flux tower GPP- and the MODIS EVI- and MCD12Q2-retrieved phenology dates were 8, 13, 34, and 16 days and 8, 19, 28, and 18 days for the start of the growing season (SOS), end of the growing season (EOS), peak of the growing season (POS), and growing season length (GSL), respectively, suggesting that MODIS EVI and MCD12Q2 were valuable products for retrieving the vegetation phenological dynamics, except for POS. The partial correlation analysis showed that temperature had a more important role in phenophase than precipitation. Rising temperatures in spring triggered earlier SOS. Significant correlations were found between EOSEVI, EOSMCD12Q2, EOSGPP, and preseason temperature (60 d prior to EOS) (p < 0.05), indicating that warming over the preseason resulted in earlier EOS. These findings help to understand the discrepancies between MODIS EVI, MCD12Q2, and flux tower GPP-based phenologies, their responses to climatic factors, and further confirm that precipitation should be taken into account to improve the accuracy of phenology models.