Responses of the terrestrial carbon cycle to drought: modeling sensitivities of the interactive nitrogen and dynamic vegetation

Drought can trigger both immediate and time-lagged responses of terrestrial ecosystems and even cause sizeable positive feedbacks to climate warming. In this study, the influences of interactive nitrogen (N) and dynamic vegetation (DV) on the response of the carbon cycle in terrestrial ecosystems of China to drought were investigated using the Community Land Model version 4.5 (CLM4.5). Model simulations from three configurations of CLM4.5 (C, carbon cycle only; CN, dynamic carbon and nitrogen cycle; CNDV, dynamic carbon and nitrogen cycle as well as dynamic vegetation) between 1961 and 2010 showed that the incorporation of a prognostic N cycle and DV into CLM4.5 reduce the predicted annual means and inter-annual variability of predicted gross primary production (GPP) and net ecosystem production (NEP), except for a slight increase in NEP for CNDV compared to CN. These model improvements resulted in better agreement with observations (7.0 PgC yr-1) of annual GPP over the terrestrial ecosystems in China for CLM45-CN (7.5 PgC yr-1) and CLM45-CNDV (7.3 PgC yr-1) than for CLM45-C (10.9 PgC yr-1). Compared to the CLM45-C, the carbon-nitrogen coupling strengthened the predicted response of GPP to drought, resulting in a higher correlation with the standardized precipitation index (SPI; rC = 0.62, rCN = 0.67), but led to a weaker sensitivity of NEP to SPI (rC = 0.51, rCN = 0.45). The CLM45-CNDV had the longest lagged responses of GPP to drought among the three configurations. These results enhance our understanding of the response of the terrestrial carbon cycle to drought.