Temporally corrected long-term satellite solar-induced fluorescence leads to improved estimation of global trends in vegetation photosynthesis during 1995–2018

Abstract

Terrestrial gross primary production (GPP) is one of the largest flux of carbon cycle and also introduces the largest uncertainties of the estimated global carbon budget. During the past decades, the emerging remotely-sensed solar-induced chlorophyll fluorescence (SIF) data have been rapidly developed as the novel proxy of GPP. Though satellite SIF datasets have been widely used for plenty of climatic and ecological studies, a lack of consistent long-term satellite SIF data has hindered its use for long-term assessments due to their temporal inconsistency induced by sensor degradation. Here, based on the time series of SIF data at the Sahara Desert as the benchmark, we present a practical approach to correct the temporal inconsistency of different satellites (GOME, SCIAMACHY and GOME-2) far-red SIF data. The temporally corrected SIF datasets were then spatially downscaled to a finer resolution of 0.05° with the light-use-efficiency (LUE) method. Then, the individual time series of SIF data from the three satellite sensors were fused together to generate a continuously long-term consistent global SIF dataset from July 1995 to December 2018 (termed as LT_SIFc*). We further validated and compared the spatial and temporal characteristics of LT_SIFc* with the ground-based long-term GPP observations from 15 flux sites, modelled global long-term GPP time series, and high-spatial resolution SIF from TROPOMI and OCO-2, respectively. Our results show that the LT_SIFc* can capture the temporal trends and inter-annual variations for both site- and global-GPP, and also the global spatial distributions of SIF at the spatial resolution of 0.05° validated with TROPOMI SIF (R2 = 0.76, RMSE = 0.11 mW m−2 nm−1 sr-1). Our temporally corrected and spatially downscaled global long-term SIF product has the potential for the better estimations of global spatial–temporal variations in GPP, and also could be helpful for the improvement in the predictions of future land carbon storage and climate change.