Improving the estimation of terrestrial gross primary productivity by downscaling global sun-induced chlorophyll fluorescence

Abstract:

The synoptic nature of satellite remote sensing makes this technique a key tool to contribute to estimating the amount of Carbon fixed by vegetation at global scale. From the various types of information that can be derived from space, the recent capacity to create global datasets of sun-induced chlorophyll fluorescence (SIF) may prove to be a game-changer. SIF is a signal emitted by the photosynthetic machinery itself that, under the illumination conditions in which it can be estimated by satellite, has been shown to be proportional to gross primary productivity (GPP). However, this relationship is dependent on vegetation types that are typically spatially mixed at the coarse spatial resolution of SIF datasets (at best 0.5°), which in turn is a consequence of the complexity of the SIF retrieval itself. This study demonstrates how 0.5° SIF derived from GOME-2 data can be downscaled to a more adequate spatial resolution of 0.05° by combining 3 explanatory biophysical variables derived from the MODIS sensor (NDVI, land surface temperature and evapotranspiration) under a semi-empirical light-use efficiency framework. The finer spatial resolution results in a cleaner signal when aggregating it per land cover type. The signal is also better correlated in time with GPP estimated from flux towers, reaching the same level of performance than global GPP products calibrated on such flux towers and driven by meteorological and remote sensing variables (other than SIF). Establishing linear relationships between SIF and flux-tower GPP at vegetation type level allows to estimate values of global terrestrial vegetation gross productivity that have different magnitude but similar temporal patterns as other GPP products. Based on downscaled SIF, the mean global GPP values over the period 2007 to 2013 are (for deciduous broadleaf and mixed forests) 13.7, (for evergreen needleleaf forests) 2.5, (for grasslands) 12.5 and (savannahs and woody savannas) 36.8 Pg of Carbon per year.