Remote Sensing Regression Model And Aridity Threshold Of Vegetation Aboveground Biomass At A Global Scale

In the context of global warming,the scope and frequency of aridity events continue to increase.Aridity has a significant impact on the structure and function of terrestrial ecosystems,therefore,a correct assessment of the aridity state of the ecosystem is significance for maintaining the stability and development of the ecosystem.Aboveground Biomass(AGB)of vegetation is the most sensitive part of terrestrial ecosystems to climate change,which can effectively reflect the health status of terrestrial ecosystems and the extent of they are affected by aridity.However,how the aboveground biomass of various vegetation types such as grassland,shrubs and forests respond to increased aridity and whether there are thresholds still needs to be systematically studied.The establishment of vegetation aboveground biomass remote sensing regression models can effectively realize the estimation and monitoring of vegetation aboveground biomass on a global scale,and help to evaluate the response mode of vegetation aboveground biomass on a global scale to drought increase.Therefore,this study sets two main contents:(1)Collect the measured AGB and remote sensing vegetation index(maximum and average value of NDVI and EVI)of grasslands,shrubs and forests vegetation types,establish linear,quadratic polynomial and exponential regression models between them,and screen the optimal regression model of each single and cross-vegetation types(that is,containing two or more types of mixed vegetation)AGB;(2)Analyze the linear or non-linear relationship between the AGB and the aridity index for each vegetation type,then use the piecewise regression to determine the threshold of the measured and simulated by the optimal remote sensing model AGB along the aridity increase.Research indicates:1.There are differences in the optimal remote sensing regression models of AGB for different vegetation types.At a global scale,the AGB optimal remote sensing regression model of cross-vegetation types is a quadratic polynomial model based on the maximum value of EVI(y=60.90-226x+1560x2,x is the max value of EVI,y is AGB,the correlation coefficient R2=0.87,p<0.001,and the root mean square error RMSE is 107.74 g m-2);The optimal remote sensing regression model of grasslands AGB is a quadratic polynomial model based on the average value of NDVI(y=65.70-181x+741x2,R2=0.63,p<0.001,RMSE is 79.71 g m-2);The optimal remote sensing regression model of shrubs AGB is a quadratic polynomial model based on the average value of EVI(y=39.40+0.30x+1350x2,R2=0.74,p<0.001,RMSE is 102.97 g m-2);The optimal remote sensing regression model of forests AGB is a linear model based on the maximum value of EVI(y=1590x-336,R2=0.88,p<0.001,RMSE is 146.03 g m-2).This optimal regression models have been cross-validated and showed good robustness,indicating that they have strong application potential.2.Both the measured and AGB simulated by the optimal remote sensing model have thresholds along the aridity increase,and they are relatively consistent.Grasslands AGB has the highest threshold along the aridity increase,and forests and crossvegetation types AGB have the lowest threshold.The order of the aridity threshold of the measured aboveground biomass of different vegetation types from high to low is: the aridity threshold of grasslands AGB(0.60)> the aridity threshold of shrubs AGB(0.29)> the aridity threshold of forests AGB(0.15)= the aridity threshold of crossvegetation types AGB(0.15).When the degree of aridity exceeds the threshold,the negative effect of increased aridity on forest and cross-vegetation types AGB is more significant.This result also proves the feasibility of using the optimal remote sensing regression model to estimate the threshold of vegetation aboveground biomass increase along aridity.The results of this study provide the remote sensing regression model for a variety of vegetation types around the world,and provide a scientific basis for the selection of the optimal remote sensing vegetation index when the regression model is constructed.At the same time,the aridity thresholds of AGB for different vegetation types have been determined,which is helpful for timely monitoring of vegetation abrupt change along the aridity increase,and has great significance to maintain the stability and development of the ecosystem.