Precipitation and carbon-water coupling jointly control the interannual variability of global land gross primary production.

Yao Zhang,Stephen Sitch,Xiangming Xiao,Xiaocui Wu,Andy Wiltshire,Atul K Jain,Joanna Joiner,Luis Guanter,Sha Zhou,Philippe Ciais,Benjamin D Stocker,Julia Nabel,Etsushi Kato,Jinwei Dong

doi:10.1038/srep39748

Abstract

Carbon uptake by terrestrial ecosystems is increasing along with the rising of atmospheric CO2 concentration. Embedded in this trend, recent studies suggested that the interannual variability (IAV) of global carbon fluxes may be dominated by semi-arid ecosystems, but the underlying mechanisms of this high variability in these specific regions are not well known. Here we derive an ensemble of gross primary production (GPP) estimates using the average of three data-driven models and eleven process-based models. These models are weighted by their spatial representativeness of the satellite-based solar-induced chlorophyll fluorescence (SIF). We then use this weighted GPP ensemble to investigate the GPP variability for different aridity regimes. We show that semi-arid regions contribute to 57% of the detrended IAV of global GPP. Moreover, in regions with higher GPP variability, GPP fluctuations are mostly controlled by precipitation and strongly coupled with evapotranspiration (ET). This higher GPP IAV in semi-arid regions is co-limited by supply (precipitation)-induced ET variability and GPP-ET coupling strength. Our results demonstrate the importance of semi-arid regions to the global terrestrial carbon cycle and posit that there will be larger GPP and ET variations in the future with changes in precipitation patterns and dryland expansion.

Highlights

The underlying mechanism is not well established
The spatial patterns of mean annual gross primary production (GPP) and interannual variability (IAV) of GPP are similar for different weight orders and model groups (Supporting Information, Figs S7–S9)
With this weighted ensemble of GPP from 14 models, we explore the linkage between the spatial patterns of GPP IAV and aridity index (Supporting Information, Table S1), i.e., the ratio of long-term annual precipitation to potential evapotranspirative demand

Summary

Results and Discussion

The annual mean and IAV of the weighted ensemble of GPP (GPPens) show very good spatial consistency with the average of means and IAVs from individual models (GPPi) and GOME-2 SIF retrievals[24] (Fig. 1). This equation only applies for regions where ET and GPP are strongly coupled on a temporal scale, i.e., high GPP variability areas as shown in the previous analysis (Fig. 3e) We calculate this βfactor by using the linear regression slopes for each pixel between the weighted GPP anomalies and ET anomalies from 12 different ET products (Supporting Information, Fig. S12), and related it with the aridity index (Fig. 4a). GPP variability patterns produced by either process-based ensemble and data-driven ensembles are similar, we note that not all models captured these IAV patterns accurately, even models with similar distributions of annual GPP With this precipitation-ET-GPP relationship being explained and the new ensemble framework developed, it provides a means to benchmark ecosystem models for water and carbon fluxes and their coupling

Materials and Methods

Author Contributions

Additional Information