Improvement of Soil Respiration Parameterization in a Dynamic Global Vegetation Model and Its Impact on the Simulation of Terrestrial Carbon Fluxes

Abstract. Soil decomposition is one of the critical processes for maintaining a terrestrial ecosystem and the global carbon cycle. The sensitivity of soil respiration (Rs) to temperature, the so-called Q10 value, is required for parameterizing the soil decomposition process and is assumed to be a constant in conventional numerical models, while realistically it is not in cases of spatiotemporal heterogeneity. This study develops a new parameterization method for determining Q10 by considering the soil respiration dependence on soil temperature and moisture obtained by multiple regression. This study further investigates the impacts of the new parameterization on the global terrestrial carbon flux. Our results show that non-uniform spatial distribution of Q10 tends to represent the dependence of the soil respiration process on heterogeneous surface vegetation type compared with the control simulation using a uniform Q10. Moreover, it tends to improve the simulation of the observed relationship between soil respiration and soil temperature and moisture, particularly over cold and dry regions. The new parameterization improves the simulation of gross primary production (GPP). It leads to a more realistic spatial distribution of GPP, particularly over high latitudes (60–80 N) where the original model has a significant underestimation bias. In addition, overestimation bias of GPP in the tropics and the midlatitudes is significantly reduced. Improvement in the spatial distribution of GPP leads to a substantial reduction of global mean bias of GPP from +9.11 to +1.68 GtC yr⁻¹ compared with the FLUXNET-MTE observation data.