| Accurate assessment of terrestrial ecosystem carbon cycle is an important basis for predicting climate change and its impact on terrestrial ecosystem.After a long period of observation and simulation,great progress has been made in the field of carbon cycle.However,due to the complexity of ecosystems and the limitations of observational and modeling methods,there is still a great deal of uncertainty in the study of terrestrial ecosystems' carbon cycle and the estimation of their spatial and temporal patterns.Therefore,how to make full use of existing multi-scale observations of the carbon cycle to identify restriction in different observation data on the model,the key parameters for typical ecosystem information,improve the accuracy of the carbon cycle count,reduce the carbon cycle and the uncertainty of the estimates of income and expenses,a reasonable objective evaluation of the carbon cycle has become the carbon cycle research field is an urgent need to solve the problem.In this study,we used long-term observational data and model data fusion method(Model-Data Fusion,MDF)to study the model parameterization and the application of the model.By using the DALEC model and the long-term carbon cycle observation data(biometric data and flux data)for parameter optimization analysis,the effects of different types of constraint data on parameter estimation and related state variables were systematically discussed.Then on this basis,the use of domestic typical subtropical forest plantation ecosystem observation data(Huitong subtropical plantation coniferous forest(HTF),Qianyanzhou subtropical plantation coniferous forest(QYZ)and Heshan subtropical broad-leaved forest(HSF)),to obtain the key parameters of typical subtropical plantation ecosystem information in China,combining with the climate conditions,and analyzed the cause of the model is the key parameter space variation.The main results of this paper are as follows:(1)the observed data of plant carbon pool significantly reduce the uncertainty of parameter estimation and model prediction,because these data provide information on the size of related carbon pool,and assimilation of soil carbon data plays an important role in reducing the uncertainty of soil carbon pool.Assimilation of multi-scale and multi-type observational data is crucial to the constraint model.When all the biometric data and flux data is used to constraint model,the simulation results the best,leaf area index(LAI),foliage carbon pool(Cf),woodiness carbon pool(Cw),fine roots carbon pool(Cr),soil carbon pool(Csom),litter carbon pool(Clit),net ecosystem carbon exchange(NEE)and soil respiration(Rs)between simulated values and observed values of root mean square error(RMSE)and mean absolute error(MAE)were 0.12 m m-2,346.82,164.21,2.10,113.20,13.52,1.44,0.66 g C m-2;0.12 m m2,215.71,111.73,1.79,90.68,10.50,1.17,0.55 g C m-2,respectively.(2)under the influence of environmental conditions and physiological characteristics of vegetation,the leaf turnover rate of QYZ and HSF was significantly higher than that of HTF.The difference of NPP(Fnf)distributed to leaves was small among the three plantation ecosystems.Three kinds of plantation ecosystem autotrophic respiration accounted for the proportion of GPP(Fg)in QYZ is the largest,the optimized parameters of mean value of 0.57,HTF next,HSF minimum,parameters of the temperature sensitive coefficient(Et)and Fg similar,QYZ is the largest,0.08.The annual change range of NEE per unit area of HTF,QYZ and HSF was-30.52?-282.85,?142.92?-653.68,-484.05?-1209.87 g C m-2 a-1,respectively,and the coefficient of variation was 0,56,0.43,0.31,respectively.NEE presented a seasonal change pattern of "double peaks"(2005-2015). |
PDF Full Text Request