Plant phenology is a key indicator of the terrestrial biosphere’s response to climate change, as well as a driver of global climate through changes in the carbon, energy and water cycles. Remote sensing observations of seasonal canopy greenness dynamics represent a valuable means to study land surface phenology (LSP) at scales relevant for comparison with regional climate information as well as ecosystem-level CO2 fluxes. We explore relationships among key LSP dates at the start and end of the season captured by three remote sensing products (i.e., NDVI: Normalized Difference Vegetation Index; PI: Phenology Index; MODIS Land Cover Dynamics Product based on the Enhanced Vegetation Index, EVI) over 19 deciduous broadleaf and mixed forest sites in the northern hemisphere for 2000–2012, and compare these estimates to estimates of start and end of photosynthesis phenology extracted from gross primary productivity (GPP) from CO2 uflux measurements. To derive phenological transition dates, we use analytical solutions of various derivatives from the fitted logistic curves. LSP dates estimated by the three remote sensing products were not equivalent and differed in their sign and magnitude of lags with photosynthesis phenology dates. NDVI-derived phenology was characterized by shorter growing seasons, while EVI prolonged it by about two weeks compared to the photosynthesis phenology season length. PI start and end of season dates more closely matched the start (r2= 0.84, RMSE = 7.61) and end (r2= 0.61, RMSE = 8.57) of photosynthesis phenology as estimated by GPP time series. PI was also found agreeing best with LSP estimates from highly spatially resolved ground digital camera observations, available for about half of the investigated FLUXNET sites. Although there were strong relationships between remotely sensed LSP and photosynthesis phenology, the relationships were not consistent across deciduous forest ecosystems implying that the vegetative and photosynthetic timing do not always follow each other in the same direction.
