Study On Parameter Optimization Of Biome-BGC Model And Estimation Of Forest Carbon Flux In Northeast China

Forests are the mainstay of terrestrial ecosystems and play an important role in mitigating the greenhouse effect and maintaining global carbon balance.Evaluation of carbon flux in forest ecosystem at regional scale is an important part of the global carbon cycle and the corresponding regional research on the global change.Process model at regional scale is an important way to analyze and predict the regional carbon balance.In addition,it is also a powerful guarantee for the sustainable development of forestry under the regional scale.Therefore,quantifying the carbon fluxes of forest ecosystems is of great significance for the study of terrestrial ecosystem carbon cycle and global carbon sources/sinks.In this study,the forest in Northeast China was taken as the research object,using remote sensing and GIS technology to obtain multi-temporal spatial information data,the flux data,remote sensing data,MT-CLIM model and Biome-BGC model are combined to construct Biome-BGC model at regional scale for forest ecosystem in Northeast China.In this study,PEST(Model-independent Parameter Estimation)and EnKF(Ensemble Kalman Filter)are used to optimize the parameters of the Biome-BGC model based on the input data,model suitability,model validation and accuracy.Through this study,we can further improve the carbon flux simulation of the model,estimate of carbon flux of forest ecosystem in Northeast China.at different spatial and temporal scales,such as year,month and day at different time scales and 1km spatial resolution.Through this study,the main conclusions are as follows:1.Parameterization of Biome-BGC model driven data:The model is driven by meteorological data,basic geographic data and vegetation physiological and ecolological parameters.The 2000-2015 daily meteorological data are selected from 0.5 ° × 0.5 °meteorological and precipitation grid data sets.The multiple time series of meteorological grid data used as the meteorological input data for Biome-BGC model are generated based on 362 meteorological grid points through the MT-CLIM model and Kriging interpolation method in the study area.The meteorological grid database of northeast regional is established.In addition,batch processing programs are developed based on ENVI/IDL platform to improve data processing efficiency and data quality.2.Parameter optimization of Biome-BGC model:The applicability of the optimization method,model improvement and operation calculation,time and cost are comprehensively considered.On site scale,the initial/state parameters of the model(N fixed parameter and maximum stomatal conductance parameter)are optimized by PEST method,while the state variables parameters of the model(the sun leaf Vmax,shade leaf Vmax,total projection LAI and decomposition rate of the total rate_scalar)are optimized by EnKF method.It is in order to achieve the complementary advantages of the two optimization methods.The programs are developed based on Visual studio2003 platform,the algorithm is coupled with the model to optimize the model parameters.Through the data assimilation,the two optimization methods not only optimize the carbon flux,but also optimize the ecosystem model parameters.On regional scale,the spatial-temporal data grid as the model input data,the site scale parameter optimization results applied to the regional model,so as to improve the simulation accuracy and the adaptability.3.Biome-BGC model simulation and operation:On site scale,the model simulates the total primary productivity(GPP)and the total ecosystem respiration(Re)of the carbon fluxes of forest ecological station in the Maoer Mountain in 2011;On regional scale,the model simulates the annual and monthly carbon fluxes of the Northeast forest ecosystem during 2000-2015,which mainly include Gross Primary Productivity(GPP),Net Primary Productivity(NPP),Net Ecosystem Productivity(NEP),Maintenance Respiration(MR),Growth respiration(GR)and Heterotrophic respiration(I IR).To realize the operation of model and the output grid simulation results of model at regional scale,the programs are developed based on ENVI/IDL platform,which uses data block,batch and multi thread parallel computing mode.It effectively improve the efficiency of regional scale model simulation and visualization.4.The temporal and spatial analysis of carbon fluxes in the forest ecosystems of Northeast China:The annual values changes during 2000 to 2015 of GPP,NPP and NEP are 334.6 to 1993.8gC·m-2·a-1,40 to 828.19 gC·m-2·a-1,and-349.78 to 314.08 gC·m-2·a-1,respectively.The annual values changes of MR,GR and HR are 72.49 to 1133.99 gC·m-2·a-1,51.39 to 247.59 gC·m-2·a-1 and 195.87 to 960.76 gC·m-2·a-1,respectively.Annual average of carbon flux relationship between GPP,NPP,MR,GR and HR in spatial distribution during 2000 to 2015:NEP is Heilongjiang Province bigger than Liaoning Province bigger than Jilin Provincc,while relationships of-other carbon fluxes are Jilin Province bigger than Jilin Province bigger than Heilongjiang Province.NEP is Great Xingan Mountains bigger than Changbai Mountains bigger than Small Xingan Mountains,MR is Changbai Mountains bigger than Great Xingan Mountains bigger than Small Xingan Mountains,relationships of other carbon fluxes are Changbai Mountains bigger than Small Xingan Mountains bigger than Great Xingan Mountains.According to the types of forest vegetation,there was no significant difference between the annual average values of carbon flux in the coniferous forest,broad-leaved forest and mixed forest in the Northeast Forest.The monthly values changes in 2011 of GPP,NPP and NEP are 0 to 335.39 gC·m-2,-54.7 to 170.19 gC·m-2 and-113.06 to 85 gC·m-2 respectively;The monthly values changes of MR,GR and HR are 0.18 to 176.95 gC·m-2,0 to 50.43 gC·m-2 and 0.52 to 153.12 gC·m-2,respectively.Monthly average of carbon flux relationship in spatial distribution between GPP,NPP,MR,GR and HR:NEP is Heilongjiang Province bigger than Liaoning Province bigger than Jilin Province,while relationships of other carbon fluxes are Liaoning Province bigger than Jilin Province bigger than Heilongjiang Province.GPP and NEP are Changbai Mountains bigger than Great Xingan Mountains bigger than Small Xingan Mountains,GR is Great Xingan Mountains bigger than Changbai Mountains bigger than Small Xingan Mountains Mountain,NPP,MR and HR is Small Xingan Mountains bigger than Changbai Mountains bigger than Great Xingan Mountains.NEP is broad-leaved forest bigger than mixed forest bigger than needle forest,MR is coniferous forest bigger than broad-leaved forest bigger than mixed forest,while relationships of other carbon fluxes are coniferous forest bigger than mixed forest bigger than broad-leaved forest.5.Analysis of carbon source/sink in the forest ecosystems of Northeast China:2000-2015 simulation results show that the annual average value of NEP of the northeast forest except in 2010,2013,is about 50 gC·m-2·a-1.The northeast forest ecosystem is in a stable state of the top community ecosystem,and its carbon sequestration capacity is relatively stable.Under the influence of climate change and disturbance factors,it will decrease the NEP value and cause the fluctuation of the interannual variation and the degree of drought determines the NEP value.2011 simulation results show that the monthly average value of NEP:It showes a single peak curve that the maximum value of NEP is 33.72 gC·m-2 in June and the minimum value of it is-24.47 gC·m-2 in April.The monthly value of NEP has the characteristics of non-growing season,which is very low,but the 4-11 month values of NEP are higher in the growing season,and there are two low values NEP at the beginning and the end of the growing month.The change of NEP was not only related to climate change and disturbance factors,but also related to the growth status of forest.There is a large fluctuation of NEP between growing season and non growing season.6.Analysis of the simulation respiration in Northeast Forest Ecosystem:According to the 2000-2015 simulation results,the annual average values of MR,GR and HR are about 265gC·m-2·a-1,150gC·m-2·a-1 and 450gC·m-2·a-1,respectively.The interannual variation trend of MR and HR of forest carbon flux in Northeast China is gradually increasing.The simulation results in 2011 show a single peak curve,which has the characteristics of low growth season value and high growth season value.The maximum value of the MR monthly average value is 59.34gC·m-2 in August and the minimum value is 2.95gC·m-2 in January;The maximum value of the GR monthly average value is 31.11 gC·m-2 in June and the minimum value of it is 0gC·m-2 in January and December;The maximum value of the HR monthly average value is 77.38gC·m-2 in July and the minimum value of it is 7.51gC·m-2 in December.The three variables relationship of the annual value and monthly value are HR bigger than MR bigger than GR.Heterotrophic respiration in northeastern forests is stronger than carbon release from autotrophic respiration.7.Spatio-temporal simulation and evaluation of carbon flux in Northeast Forest Ecosystem:On site scale,comparison of the simulation results with the flux observation data of the forest ecological station is the GPP and Re of the R2 are improved by 10.67%and 10.59%,RMSE is 2.53 and 0.55 which are decreased by 22.87%and 57.03%,respectively,the total of GPP and Re in the growing season increased by 17%and 21.72%compared with the default model.On regional scale,the simulation results are compared with sample plot data of forest inventory and MODIS remote sensing data products,and the results of different scholars’ model estimation.The verification shows that the Biome-BGC model simulation achieves better results.Therefore,it is feasible and effective to demonstrate that the application of Biome-BGC model on regional scale in the northeastern forest.It will be helpful to realize the integrated application of flux observation,remote sensing data and ecological model and provide a reference for the study of spatial and temporal distribution patterns of forest carbon source/sink.