Land surface phenology detections from multi-source remote sensing indices capturing canopy photosynthesis phenology across major land cover types in the Northern Hemisphere

Land surface phenology, which records the start of growing season (SOS) and the end of growing season (EOS), plays an essential part in reflecting plant photosynthesis and the response of carbon cycle in terrestrial ecosystems to climate change. Significant advances have been made toward tracking vegetation responses to climate variability based on land surface phenology derived from satellite remote sensing information. However, the advantages and disadvantages of single remote sensing index in estimating land surface phenology across major land cover types has not been well documented, which hindered our ability to better understand the impact of climate variability on plant phenology at large scales. In our study, four remote sensing indices, including solarinduced chlorophyll fluorescence (SIF), leaf area index (LAI), normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) based on 66 eddy flux tower sites in the Northern Hemisphere during the period of 2007-2014, were integrated to estimate land surface phenology across five land cover types, including evergreen needle-leaf forests, deciduous broadleaf forests, mixed forests, grasslands and croplands. The phenology extracted from gross primary production (GPP) from eddy covariance measurements was treated as real canopy photosynthesis phenology to verify the estimates of phenology transitions based on remote sensing indices. Results showed that all four remote sensing indices can capture land surface phenology, but showed different ability within land cover types. In details, phenology derived from LAI and SIF in three types of forests appeared to have good relationships with canopy photosynthesis phenology based on GPP, while phenology based on EVI or NDVI was close to GPP based phenology at grasslands and croplands sites. Meanwhile, the integration of four remote sensing indices could estimate land surface phenology more comparable to canopy photosynthesis phenology than a single remote sensing index for most sites. Furthermore, SOS was affected primarily by shortwave radiation, while EOS was regulated by a combination of different climatic variables in the Northern Hemisphere. The integration of remote sensing indices phenology could improve the capacity of estimating phenology transitions, which help us to better understand the impacts of climatic variables on land surface phenology and vegetation dynamics in future climate change.