Predicting deciduous forest carbon uptake phenology by upscaling FLUXNET measurements using remote sensing data

Terrestrial ecosystems are highly sensitive to climatic changes in early and late growing seasons. Land surface phenology (LSP), the study of the timing of recurring seasonal pattern of variation in vegetated land surfaces observed from synoptic sensors, has thus received much attention due to its role as a surrogate in detecting the impact of climate change. Although several studies have been conducted on the growing season LSP, studies on the net carbon uptake phenology (CUP) defined as the detrended zero-crossing timing of net ecosystem productivity from a source to a sink in spring and vice versa in autumn, have been scarce. Here we present a CUP determination approach using the commonly available remote sensing data in four selected temperate and boreal deciduous forest CO2 flux tower sites. We test a hypothesis that the mean monthly surface temperature and LSP derived from remote sensing observations explain the CUP both in spring and autumn seasons. Our approach predicts the observed CUP in spring and autumn within 8 day mean errors, equivalent to the temporal resolution of the 8-day composite remote sensing dataset used in this study for the four flux tower sites. The results from this study will have a large implication for global change studies with increasing amount of valuable remote sensing data to be used for monitoring CUP beyond the footprints of CO2 flux towers.