Including soil water stress in process-based ecosystem models by scaling r down maximum carboxylation rate using accumulated soil water deficit

摘要

The impacts of soil water stress on photosynthesis and stomatal conductance have not been uniformly parameterized in terrestrial ecosystem models. This study firstly quantifies diurnal variations in canopy conductance (g_(c,w)) during full-leaf periods from eddy covariance flux data at four flux sites by inverting the Penman-Monteith equation, Then, the Ball-Woodrow-Berry (BWB) slopes were derived and compared under contrasting water stress conditions by linear regression of g_(c,w) and gross primaryproductivity (GPP) derived from EC measurements. Finally, the response of the leaf maximum carboxylation rate to accumulated soil water deficit (ASWD) was explored to develop a better scheme of soil water stress. Our results show: (1) The thresholds ofrelatively available soil moisture under which soil water stress occurs were 0.575, 0.885, 0.495 and 0.653 for the tropical savanna site (AU-How), the Mediterranean forest site (IT-Col), the Mediterranean grassland site (US-Var) and the boreal forest site (CA-Oas), respectively, derived from the logistic functions of fitting g_(c,w) to relatively available soil moisture; (2) similar to previous work, we found that ensemble average g_(c,w) of dry periods were lower than those of wet periods at the studied sites, and that BWB slopes did not change significantly during droughts, indicating that BWB slopes may be conserved under prolonged drought; and (3) EC-derived GPP gradually decreased with the increase of ASWD, which can be well captured by a V_(mr)-ASWD scheme developed in this study. In sum, the V_(mr)-ASWD scheme would increase the accuracy of GPP simulations of ecosystem models. This study suggests that the change of BWB slopes under prolonged drought is not justified according to the experimental data examined while adjusting V_(cmax) with ASWD accounts for the change in leaf physiology due to prolonged drought and is computationally feasible and efficient.