Improvement of predicting ecosystem productivity by modifying carbon–water–nitrogen coupling processes in a temperate grassland

Abstract

AbstractAimsPrediction of changes in ecosystem gross primary productivity (GPP) in response to climatic variability is a core mission in the field of global change ecology. However, it remains a big challenge for the model community to reproduce the interannual variation (IAV) of GPP in arid ecosystems. Accurate estimates of soil water content (SWC) and GPP sensitivity to SWC are the two most critical aspects for predicting the IAV of GPP in arid ecosystems.MethodsWe took a widely used model Biome-BGC as an example, to improve the model performances in a temperate grassland ecosystem. Firstly, we updated the estimation of SWC by modifying modules of evapotranspiration, SWC vertical profile and field capacity. Secondly, we modified the function of controlling water–nitrogen relation, which regulates the GPP–SWC sensitivity.Important FindingsThe original Biome-BGC overestimated the SWC and underestimated the IAV of GPP sensitivity, resulting in lower IAV of GPP than the observations, e.g. it largely underestimated the reduction of GPP in drought years. In comparison, the modified model accurately reproduced the observed seasonal and IAVs in SWC, especially in the surface layer. Simulated GPP–SWC sensitivity was also enhanced and became closer to the observations by optimizing parameter controlling nitrogen mineralization. Consequently, the model’s capability of reproducing IAV of GPP has been largely improved by the modifications. Our results demonstrate that SWC in the surface layer and the consequent effects on nitrogen availability should be among the first considerations for accurate modeling IAV of GPP in arid ecosystems.