The Research On Evaluating GPP Based On Multi-source Data And Effects Of LUCC On Carbon Stocks In China

In recent decades,rapid economic development and a sharp increase in population have led to significant changes in land use types in China,including rapid urbanization,agricultural development,as well as a series of regional development and ecological protection policies such as"afforestation","returning farmland to forests",and"natural forest conservation projects"aimed at improving the ecological environment.These changes have had an important impact on the national ecological and environmental conditions and regional climate change.Land use and cover change(LUCC)is one of the main factors influencing terrestrial carbon cycling and significantly affects regional carbon and water cycles,energy transmission,climate change,and biodiversity.Therefore,quantitatively assessing the carbon changes caused by land use change is crucial for balancing regional carbon budgets and understanding the impact of human activities on the ecological environment.Bookkeeping models are currently the mainstream method for quantitatively evaluating the impact of land use change on carbon stocks.However,this model mainly relies on empirical parameters and cannot be compared with observational data,leading to significant uncertainties.This study aims to integrate the bookkeeping model with a process-based dynamic model to accurately quantify the impact of land use change on the carbon balance of terrestrial ecosystems.In addition,Gross Primary Production(GPP),which is the amount of carbon absorbed by vegetation through photosynthesis,is the largest carbon flux between land ecosystems and the atmosphere and is an essential factor in determining the carbon balance of ecosystems and reflecting their response to global changes.Accurately estimating the GPP of terrestrial ecosystems is the basis for quantitatively evaluating carbon balance,and various light-use models have been developed for this purpose.However,the variables and algorithms used to describe environmental limiting factors differ greatly among these models,and exploring the use of machine learning methods and different variable combinations in light-use models to improve the accuracy of GPP estimation is crucial.The study focused on This study focuses on Chinese terrestrial ecosystems and utilizes the Chinese Flux Observation and Research Network(China FLUX)as research sites.The study uses meteorological data,remote sensing data,flux observation data,and land use data to construct an RFR-LUE model based on the light use efficiency(LUE)model and random forest regression(RFR)model.The study investigates the effects of different variables and time scales on GPP estimation and simulates long-term GPP data from the model as validation data for the process-based 3-PGS model.Using spatial overlay analysis in GIS technology,the study analyzes land use changes from 2000 to 2018 and quantitatively evaluates the impact of land use change on carbon storage in Chinese terrestrial ecosystems by coupling the 3-PGS model and the Bookkeeping empirical model.(1)To address the significant uncertainty of the LUE model in estimating GPP,this study constructs an RFR-LUE model by combining the variables of the LUE model with the random forest regression model.Results show that the RFR-LUE model can accurately simulate the temporal changes and amplitude of GPP.The time scale of the input data significantly affects the accuracy of the model,with an improvement in model accuracy from a daily time scale to a monthly time scale,with the average R~2 value increasing from 0.81 to 0.90.The variable importance of the RFR model indicates that vegetation index and temperature are important variables in the RFR-LUE model at most sites.Compared with GPP products estimated by four other models,the GPP predicted by the RFR-LUE model matches well with the flux observation GPP and can serve as an effective method for simulating GPP changes.(2)To address the problem that carbon storage data estimated by the Bookkeeping model cannot be compared with actual data,this study constructs a carbon quantification model for land use change by combining the process-based 3-PGS model with the Bookkeeping model.The annual GPP values of the flux observation site estimated by the RFR-LUE model serve as the validation data for the intermediate results of the 3-PGS model.Results show that the 3-PGS model estimates annual total GPP highly correlated with the RFR-LUE model,with R~2ranging from 0.70 to 0.83.The annual mean GPP at most sites is lower than the estimated value of the RFR-LUE model,within a reasonable range.The spatial distribution of GPP at the national scale estimated by the 3-PGS model shows that the GPP of terrestrial ecosystems gradually decreases from the southeast coast to the northwest.In addition,using carbon density observation data as validation,the results show that the model’s output is relatively accurate(R~2=0.62),with the overall carbon density value estimated by the model being higher than the observed value.(3)Based on land use data from five periods between 2000 and 2018,the transfer matrix and trajectory analysis method were used to analyze the process and reason of land use change.The results showed that land use change between 2000 and 2018 exhibits different spatial and temporal characteristics in different regions of China.The total area of arable land decreased,mainly in central and eastern regions.Due to the implementation of the Grain-for-Green and afforestation projects in the southwest and north China,the area of forest land increased.The grassland in the northwest region decreased significantly,especially in the agricultural and pastoral areas,mainly degraded into unused land.The construction land showed a significant expansion trend,mainly in the southeast and central regions.Trajectory analysis of land use change shows that both natural evolution and human interference are the main driving forces of land use change,but their impact on the area of land use change varies in different regions.There is a large area of human interference type of land use change in the eastern and southern regions,while land use change in the western region is mainly affected by natural factors such as climate change and water shortage.Under the influence of land use change,the carbon storage increased by 1.32 Pg from 2000 to 2018,which the carbon storage increased by about1.08 Pg due to Grain-for-Green,afforestation,and reforestation.