Evaluation of the Community Land Model simulated carbon and water fluxes against observations over ChinaFLUX sites

Tuesday, 16 December 2014: 4:30 PM
Li Zhang1, Jiafu Mao2, Xiaoying Shi2, Daniel M Ricciuto2, Honglin He1, Peter E Thornton2, Guirui Yu1, Shijie Han3, Yingnian Li4, Junhua Yan5, Yanbin Hao6 and Huimin Wang1, (1)Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China, (2)Oak Ridge National Laboratory, Oak Ridge, TN, United States, (3)Institute of Applied Ecology, Chinese Academy of Sciences, Shenyang, China, (4)Northwest Institute of Plateau Biology, Chinese Academy of Sciences, Xining, China, (5)South China Botanical Garden, Chinese Academy of Sciences, Guangzhou, China, (6)University of Chinese Academy of Sciences, Beijing, China
The Community Land Model (CLM) is an advanced process-based land surface model that simulates the complicated carbon, water vapor and energy exchanges between the terrestrial ecosystem and the atmosphere at various spatial-temporal scales. We for the first time use eddy-covariance observations of CO2 and water vapor exchange and soil respiration measurements at five representative Chinese Terrestrial Ecosystem Flux Observational Network (ChinaFLUX) tower sites to systematically evaluate the latest versions of CLM, the CLM4.0 and CLM4.5, and comprehensively examine the similarities and differences between the observational and simulated results. The CLM4.5 underestimates annual carbon sink at three forest sites and one alpine grassland site but overestimates the carbon sink at a semi-arid grassland site. The underestimation in annual carbon sink at a deciduous dominated forest site is resulted from underestimated daytime carbon sequestration in summer and overestimated nighttime carbon emission in spring and autumn. Compared with the CLM4.0, the bias of annual Gross Primary Production (GPP) is reduced by 24% and 28% in CLM4.5 at two subtropical forest sites. However, CLM4.5 still has a large positive bias in annual GPP. The improvement in NEE is limited, although the bias of soil respiration decreases by 16%-43% at three forest sites. The CLM4.5 has lower soil water content in dry season than this simulated by the CLM4.0 at two grassland sites. These lead to the significant drop in leaf area index and GPP, and the increase in respiration for the CLM4.5. The new fire parameterization in CLM4.5 causes incorrect fire estimation at Changbaishan forest site, which results in unexpected underestimation of NEE, vegetation carbon, and soil organic carbon by 46%, 95%, and 87%, respectively. Our study with the ChinaFLUX sites indicates a significant improvement of the CLM4.5 than the CLM4, and suggests further developments on the parameterization of seasonal GPP and respiration, which depend tightly on the better representation of seasonal water condition and the partitioning of net radiation between sensible and heat fluxes.