Structural uncertainty in model-simulated trends of global gross primary production

Accurate representation of the effect of drought on terrestrial vegetation functioning is important for understanding the interannual variability in global Gross Primary Production(GPP) and for projecting carbon sequestration potential by vegetation. Drought effect is usually modeled as a function of Vapor Pressure Deficit (VPD) and/or soil moisture. Global warming is likely to accelerate increasing trend in VPD, while a relatively stable precipitation is predicted. This difference in projections between VPD and precipitation can cause serious discrepancies in vegetation behavior depending on how the ecosystem models represent the drought effect. In this study, we scrutinized the model responses to drought using the 30-year record of GIMMS 3G dataset (1982-2010). A diagnostic ecosystem model, Terrestrial Observation and Prediction System (TOPS), was used to estimate global GPP from 1982 to 2009 with 9 different experimental simulations. The control run of global GPP increased until around 2000, but stayed flat after 2000. Among the simulations with single climate constraint, only the VPD-driven simulation showed a decrease in 2000s, while the other scenarios simulated an increase in GPP. These different responses in 2000s can be attributed to the difference in the representation of water stress in models, i.e. using VPD and/or precipitation. When we compared the trend of simulated GPP with CO2 growth rate, VPD-driven model had the highest correlation with CO2 growth rate. However, spatial map of trend in simulated GPP using GIMMS 3G data showed more consistent with the GPP driven by soil moisture than the GPP driven by VPD. Thus, the GPP driven by soil moisture is close to satellite observations in TOPS model, and high correlation of VPD-driven simulation with CO2 growth rate can be attributed to spurious correlation that is likely induced by the previously reported high correlations between CO2 growth rate and temperature variability.