The Uncertainty Of Simulation Of Carbon Storage In Terrestrial Ecosystem And Its Causes Traceability

The terrestrial ecosystem is an important reservoir of the global carbon cycle.The total amount of carbon stored in it is about four times that of the atmosphere,so it plays a vital role in the feedback relationship between climate change and the carbon cycle.Earth system models based on physical and biogeochemical processes are an important means to study the feedback effects of climate change and terrestrial ecosystems.However,the simulations of global terrestrial carbon storage by different earth system models are quite different.Therefore,how to trace the causes of differences in terrestrial carbon storage simulations in the Earth system model is a difficult problem in global change ecology.Existing studies have shown that the uncertainty of terrestrial carbon storage simulations of earth system models mainly comes from the structural differences between models.Then,the development of new traceability analysis tools to quickly quantify the difference in the simulation of terrestrial carbon storage by models of different structures has become a key task in reducing the uncertainty of future climate change predictions.In response to the above-mentioned needs,this study firstly analyzed the difference characteristics of current mainstream terrestrial carbon storage process models in estimating carbon storage based on the Coupled Model Intercomparison Project Phase 6(CMIP6).Then,using a traceability analysis method and a new distributed data collaborative processing framework,a new generation of model evaluation tools(Traceability analysis for model evaluation,Trace Me)was developed.Based on the Trace Me tool,this paper quantified and traced the sources of uncertainty of terrestrial carbon storage in the CMIP6 model in the simulated historical period and different future scenarios.Finally,this paper compared the first-order linear kinetic model commonly used in the Earth system models and the emerging nonlinear microbial mechanism model in recent years and evaluated the impact of the improvement of the model structure on the terrestrial carbon storage simulation,which was a further step of the terrestrial carbon storage process model.Improvement and development provided new ideas and basis.The main results are as follows:(1)Through the comparative analysis of the carbon storage simulated by 16CMIP6 models,this study found that the current terrestrial carbon cycle process models had a large uncertainty in the simulation of terrestrial carbon storage,and the difference among the models reached 4.9 times.Compared with the simulation of vegetation carbon storage,the simulation difference of soil carbon storage between models was the main source of uncertainty in terrestrial carbon storage simulations.Spatially,the simulation uncertainty of soil carbon storage in the northern high latitudes was the main source of uncertainty in the simulation of carbon storage in the terrestrial carbon cycle process models.(2)Aiming at the shortcomings of traditional model evaluation in the traceability function,this paper built a new model evaluation tool: Trace Me based on the traceability analysis method and the distributed collaborative data processing framework.This tool took the advantage of the traceability analysis method in tracing the source of uncertainty in terrestrial ecosystem carbon storage simulation,and systematically decomposed carbon storage into the ecosystem’s ability to store carbon(carbon storage capacity)and how much carbon could still be stored at present The potential of carbon storage(carbon storage potential)was then decomposed into traceable variables such as total carbon input(total primary productivity),carbon use efficiency,baseline carbon residence time,and environmental factors.At the same time,with the help of variance decomposition method,quantified their contribution to the uncertainty of terrestrial ecosystem carbon storage simulation.These variables could explain which specific carbon cycle process the uncertainty of the model simulates terrestrial carbon storage came from.Combined with the relevant mechanisms of the terrestrial carbon cycle,the evaluation results could provide more effective suggestions for the improvement of the model.(3)Analyzed the simulation uncertainty of 16 CMIP6 models on terrestrial carbon storage from 1850 to 2014,and traced the source of the uncertainty of terrestrial carbon storage simulation among models based on the Trace Me tool.This paper found that the baseline carbon residence time of soil organic carbon was the main source of uncertainty in the simulations of terrestrial carbon storage among models,with a contribution of 75.7%.In terms of time changes,the difference in terrestrial carbon storage potential among models had increased the contribution of the uncertainty of carbon storage simulation,from 2.1% in 1851 to 1860 to 21.5% in 2005 to 2014.In terms of spatial variability,gross primary productivity(GPP),carbon storage potential,and baseline carbon residence time were the main dominant factors for the uncertainty of carbon storage simulations,accounting for 41.5%,23.3%,and 34.7%,respectively.Moreover,for different ecosystems,the proportions of the dominant factors for the uncertainty of carbon storage simulations were quite different.(4)Under future scenarios(Shared Socioeconomic Pathways,SSP1-2.6,SSP2-4.5,SSP3-7.0,and SSP5-8.5),the traceability analysis of the uncertainty of carbon storage simulations by 13 CMIP6 models showed that soil baseline carbon residence time was the main source of uncertainty in terrestrial carbon storage simulations among models,contributing 63.4%,60.6%,54.9%,and 45.5% under the four future scenarios,respectively.However,as the simulation time progressed,the contribution of different traceable variables to the simulation uncertainty of terrestrial carbon storage had changed,mainly as follows: By the end of this century,as the forcing increases(from SSP1-2.6 to SSP5-8.5),the contribution of net primary productivity(NPP)and carbon storage potential to the simulation uncertainty of terrestrial carbon storage had increased significantly,from 12.9% and 6.3% to 22.1% and 22.0%,respectively.In terms of spatial distribution,the uncertainty of GPP’s simulation of terrestrial carbon storage in the area near the equator had increased significantly,while the contribution of baseline carbon residence time had decreased.In different ecosystems,the evergreen deciduous forests under high forcing scenarios(SSP5-8.5),the simulation of terrestrial carbon storage by carbon storage potential increased significantly increased from 17.2%in 2015 to 32.6 in 2100 %,and other ecosystems did not have obvious trends.(5)Aiming at the question of whether the development of the future model structure could effectively improve the accuracy of terrestrial carbon storage simulation,this paper compared and analyzed the performance of the microbial mechanism model and the conventional first-order linear kinetic model in simulating the change of soil carbon storage in a semi-arid grassland.The results showed that the current microbial mechanism model structure could not significantly improve the simulation of carbon storage by terrestrial carbon cycle process models.There is still a big difference in the simulation of soil carbon residence time between the two types of model structures.Whether the microbial mechanism model could reduce the simulated uncertainty of the carbon residence time among the traditional first-order linear kinetics models was still controversial,and more ite-based observation data for evaluation and parameterization to improve its simulation performance of carbon storage.In addition,the expression of environmental factors and their parameterization schemes in the model significantly affected the model’s performance in simulating changes in soil carbon storage and soil carbon residence time.The development of terrestrial carbon cycle models required more consideration of the characterization of environmental factors.Overall,this paper built a new model evaluation tool to solve the problem of large uncertainties in simulating carbon storage among terrestrial carbon cycle process models.Trace Me innovatively integrated the traceability method into the model evaluation,which had strong application value in quantifying and tracing the source of uncertainty in the current model simulation of carbon storage.Finally,the performances of the new terrestrial carbon storage process model structure(microbial driven model)and the conventional model structure in simulating carbon storage were compared and analyzed.This paper emphasized the importance of the relevant ecological processes and elements(carbon allocation,carbon turnover,soil properties,etc.)of the baseline carbon residence time of the soil,the dynamic changes of non-equilibrium carbon storage,and the importance of diversified spatial data in the development of terrestrial carbon cycle process models,and provided a strong basis for the improvement of the models.