Simulation Of Vegetation GPP By Combining Flux Observation Site Data And Light Energy Efficiency Model

Carbon cycle is an important way of material exchange in terrestrial ecosystem.Vegetation Gross Primary Productivity(GPP)is an important link in the study of carbon cycle.Under the background of global climate change,how to quantify the carbon flux of terrestrial ecosystems is of great significance.The flux tower site can provide regional GPP observation capacity,and the satellite remote sensing technology can provide global GPP observation capacity,both of which provide technical means for GPP observation.In recent decades,the global warming caused by the emission of carbon dioxide has caused great harm to the growth of plants and animals and human life.How to accurately understand and master the global carbon cycle process is a major problem facing scientists.Remote sensing model estimation can fit GPP more accurately.However,it is faced with such problems as large spatial scale of remote sensing but low accuracy,differences in remote sensing estimation of different vegetation types,and the impact of extreme weather on vegetation productivity under climate change.Therefore,how to fit GPP more accurately is the focus of current research.Based on precipitation,shortwave radiation,temperature and total primary productivity data,MODIS surface reflectance products,PAR and potential evapotranspiration products,GPSIF products and SPEI data from 74 flux tower observation stations in the northern hemisphere,The influence factors of GPP were analyzed and calculated by the methods of correlation analysis,structural equation model and nonlinear regression,etc.,which were used to fit GPP.The model fitting accuracy of GPP of different types of vegetation,the influence of different types of remote sensing products on the accuracy of LUE model,and the comparison of the accuracy of LUE model before and after the improvement by changing the threshold value of parameters in the model were deeply explored.The main conclusions are as follows:(1)Vegetation GPP could be affected by climate factors,and different types of vegetation GPP were affected by different climate factors.Through Pearson correlation analysis and structural equation model,it was found that temperature,water content and light were all important factors affecting the GPP of vegetation.Among them,temperature was an important influencing factor for all types of vegetation,especially for grassland and shrub.It does not directly affect the GPP of grassland and shrub,but indirectly affects the GPP of grassland and shrub by affecting other factors.Water content was the main influencing factor in deciduous broad-leaved forest,farmland and wetland.light was the main influencing factor in farmland,grassland,shrub and mixed forest.In addition,evapotranspiration was found to be an important factor affecting the GPP of all vegetation types except shrub types.(2)The FPAR parameters in the light light use efficiency model could be calculated by using EVI,SIF and MODIS FPAR.The GPP fitting accuracy of the light light use efficiency model obtained by different calculation methods was EVI,SIF and MODIS FPAR in descending order.Using EVI data,the GPP fitting accuracy of cropland,grassland and shrub type vegetation was the highest.(3)Using S-G filtering method to simulate vegetation GPP after processing remote sensing data,different vegetation types could be improved to different degrees,which could make the fitting results more accurate.The increase was the highest in evergreen broad-leaved forest and evergreen needleleaved forest,and the lowest in wetland and farmland.(4)Drought did affect vegetation GPP.The accuracy of VPMSW model was higher for cropland,deciduous broad-leaved forest,evergreen needleleaved forest,mixed forest and shrub under the influence of drought.Drought stress caused by extreme weather had different effects on GPP simulation of different vegetation types.The GPP of deciduous broad-leaved forest,evergreen needleleaved forest and grassland would be affected by short-term extreme drought,while cropland,shrub,mixed forest and wetland would not be affected by short-term extreme drought.Evergreen broad-leaved forest could adapt to short-term extreme drought.(5)After improving the temperature parameters,the accuracy of GPP fitting of grassland and shrub was improved by VPM model,and the accuracy of GPP fitting of grassland was improved greatly.Temperature had only an important indirect effect on grassland and shrub GPP,not a direct effect.According to the estimation of vegetation GPP by VPM model,it was found that grassland was under obvious low temperature stress.Therefore,adjusting the temperature parameters of the model to improve the fitting accuracy of grassland GPP could be greatly improved.Because shrubs were not subjected to obvious low temperature stress,only the optimal temperature obtained by nonlinear fitting method fluctuates greatly.After adjusting temperature parameters,the accuracy of model fitting of shrub GPP was less improved.(6)Compared with the light use efficiency model,the GPP fitted by the generalized regression neural network method could get more accurate results,except for cropland type.It was also found that evapotranspiration was an important factor affecting GPP.GRNN was used to fit GPP with different FPAR parameters,and the results of SIF and EVI were better than MODIS FPAR.GRNN was based on more accurate climate factor and remote sensing factor fitting GPP,which could obtain higher fitting results.