An improved hydro-biogeochemical model (CNMM-DNDC V6.0) for simulating dynamical forest-atmosphere exchanges of carbon and evapotranspiration at typical sites subject to subtropical and temperate monsoon climates in eastern Asia

Abstract. Carbon exchange between forest ecosystems and the atmosphere play an important role in global carbon cycle, which is difficult to be accurately quantified due to the large uncertainties in scaling up site-scale observations or filling-up measurement gaps. A process-oriented model equipped with comprehensive processes to explicitly simulating coupled carbon, nitrogen and water cycling, is hypothesized to reduce the uncertainties in quantification of forest carbon fluxes. To test this hypothesis, the Catchment Nutrient Management Model – DeNitrification-DeComposition (CNMM-DNDC), as a hydro-biogeochemical model that dynamically couples the carbon, nitrogen, phosphorous and water cycling processes, was updated in this study by incorporating a new forest growth module derived from the Biome-BGC model and validating the updates using multiple-year continuous observations of carbon and water fluxes at the site scale. The updated model has improved the processes of photosynthesis, litter decomposition, allocation, respiration and mortality to more effectively capture the transformation and transportation of nutrients in plant-soil-water continuum. The observed gross primary productivity (GPP), ecosystem respiration (ER), net ecosystem carbon dioxide exchange (NEE) and evapotranspiration (ET) of three typical forest sites subject to subtropical and temperate climates in eastern Asia (2003–2010) were used for the model calibration and validation. Compared with the original model in validation, the updated model showed significant improvements in simulating the daily dynamics and inter-annual variations of each variable, with the NRMSE values decreased by 46 % and 54 %, 65 % and 37 %, 4 % and -6 %, and 38 % and -3 % for GPP, ER, NEE, and ET on daily and annual scales, respectively. The comparable performances of both model versions for annual NEE emphasizes the importance of validating each component of carbon fluxes to avoid the offsetting of model errors. The canopy average specific leaf area, fraction of leaf nitrogen in Rubisco, annual leaf and fine root turnover fraction, maximum stomatal conductance and the ratio of carbon to nitrogen in leaves and fine roots were identified as the sensitive eco-physiological parameters affecting the simulations of GPP and ER. In addition, the meteorological variables of solar radiation, humidity and air temperature also showed strong influences on the simulated GPP and ER. The relatively satisfactory performances demonstrated that the modified model has the ability to capture the daily dynamics and inter-annual variations of carbon fluxes for forests in temperate and subtropical zones, which is essential for estimating the emissions of greenhouse gases at the regional or global scales.