Evaluation And Improvement Of Diffuse Radiation Effect Using Light Use Efficiency Model

Gross Primary Productivity（GPP）is the amount of organic matter produced by vegetation per unit time and per unit area.Accurately assessing the spatiotemporal characteristics of GPP in terrestrial ecosystems is essential for understanding the impact of the global carbon cycle processes and achieving the "double carbon" goal.Diffuse radiation is an important environmental factor.Variations in quantity and quality of diffuse radiation will affect the GPP assessments.Although many studies have been conducted to analyze the diffuse radiation effects on terrestrial GPP,various expression forms of diffuse radiation effects on vegetation carbon cycle in the LUE models can lead to differences in GPP assessments.Therefore,it is necessary to quantify the biases of these models for diffuse radiation effect.First,we calculated the bias of diffuse radiation effects between four LUE(（VPM,TEC,EC-LUE and DTEC models）models and FLUXNET sites.Second,we calibrated and validated the relationship between diffuse radiation and the maximum light use efficiency of the sunlit and shaded leaves(ε_msu、ε_msh).Then,we proposed a new twoleaf LUE model that considers the impact of diffuse radiation on sunlit and shaded leaves（Diffuse radiation Two-leaf Light Use Efficiency,DTL-LUE）.Finally,we used the DTLLUE model to analyze the spatiotemporal distribution of GPP in China from 2000 to 2020.The research contents and conclusions are as follows:（1）We used the radiation use efficiency（RUE）to represent the diffuse radiation effect in four LUE models（TEC model;VPM model;EC-LUE model;DTEC model）.The bias analysis of RUE between models and FLUXNET sites was carried out on 100 forest FLUXNET sites,and the variation of RUE with different diffuse radiation fractions was explored.Our conclusions are as follows:The big-leaf LUE model（TEC,VPM model）can only simulate the diffuse radiation effects under low diffuse radiation fraction conditions（below 0.3）and the two-leaf model can obtain the enhanced RUE by the diffuse radiation in most bins of diffuse radiation fraction.However,these LUE models simulated RUE still have a larger bias compared with the observed RUE from FLUXNET.Therefore,it is essential to further optimize the accuracy of LUE models in evaluating the diffuse radiation effects.The consideration of diffuse radiation effects on the canopy maximum light use efficiency(LUE_max)might be useful for improving the model.（2）Combining the bias analysis results of the diffuse radiation effect of the four LUE models,we found that incorporating the diffuse radiation fractions in the calculation of the maximum light use efficiency(LUE_max)of canopy can greatly improve the model evaluation accuracy.Therefore,a DTL-LUE model was developed to distinguish the effect of direct radiation and diffuse radiation on the maximum light use efficiency of the sunlit and shaded leaves(ε_msu、ε_msh).We calibrate and validate the DTL-LUE model with data from 181 FLUXNET sites covering 9 vegetation types.The verification results show that under the nine vegetation types,the ε_msu and ε_msh obtained by the improved DTL-LUE model are 0.60～1.19 gC·MJ^-1,1.72～3.31 gC·MJ^-1,respectively.ε_msh is 2.5-3.5 times ofε_msu.The RUE of each vegetation type calculated by FLUXNET showed a rising trend with the increase of diffuse radiation fractions.The RUE obtained by the DTL-LUE model is more consistent with the tread of FLUXNET.In addition,comparing the mean bias of RUE between the three two-leaf models（EC-LUE、DTEC、DTL-LUE model）and FLUXNET,the average bias of RUE simulated by the DTL-LUE model is lower than the other two-leaf LUE models.The average RUE bias of RUE was less than 0.45 gC·MJ^-1 under the nine vegetation types.Compared with the DTEC and EC-LUE models,the DTL-LUE model simulated the cumulative bias of forest RUE by 90～100 gC · MJ^-1,250～500 gC · MJ^-1,respectively.（3）Combined with leaf area index（LAI）from MODIS,direct and diffuse radiation from CERES,air temperature and dew point temperature from ECMWF and vegetation type from MOD 12,the DTL-LUE GPP products in China from 2000 to 2020 were driven.The GPP product is on a daily scale with a spatial resolution of 1°× 1°.We found that the GPP in China from 2000 to 2020 showed an increasing trend from northwest to southeast,and the annual average value of GPP was roughly 0-2800 gC · MJ^-1.In the past 20 years,the GPP in China has generally shown an upward trend.The area of the rising area accounts for 73%of the total land area,and the average annual increase of GPP is 0～64 gC · MJ^-1.The area of decline area accounted for about 5.2%of the total land area,and the average annual decline in GPP is 9.6 gC · MJ^-1.The GPP in China has generally shown an exponential growth trend,with an annual growth rate of 13 gC ·MJ^-1 in 2000-2020.The GPP of forest has increased most significantly,followed by is cropland,and the GPP of grassland has the slowest increase.