耦合叶肉导度的陆面过程模型最大叶肉导度参数的敏感性分析

陆面过程模型添加叶肉导度能有效改善模型模拟的CO₂施肥效应精度，但叶肉导度模拟受最大叶肉导度参数取值的影响，优化模型中最大叶肉导度参数是改进陆面过程模型叶肉导度和CO₂施肥效应模拟的重要途径。以EALCO（Ecological Assimilation of Land and Climate Observations）模型为例添加叶肉导度，通过人为改变最大叶肉导度值的取值，分析模型输出结果对最大叶肉导度的响应，揭示最大叶肉导度参数在模型中的敏感性，并与已有研究结果或观测数据比较，探讨耦合叶肉导度的陆面过程模型最大叶肉导度参数优化的途径。模拟试验以美国哈佛森林典型温带落叶阔叶林生态监测站（US-Ha1 site，Harvard Forest Environmental Monitoring site）数据为驱动。结果显示：（1）随最大叶肉导度增加，总初级生产力（GPP， Gross Primary Production）模拟精度增加，但最大叶肉导度取值大于1.0 mol m^-2 s^-1后模拟精度改善有限，最大叶肉导度小于1.0 mol m^-2 s^-1时GPP模拟精度对最大叶肉导度变化响应敏感；（2）证实了叶肉导度与气孔导度之间存在明显线性关系，最大叶肉导度取值的变化能明显影响这种线性关系的斜率。当最大叶肉导度取值从0.5 mol m^-2 s^-1增加到1.2 mol m^-2 s^-1时，气孔导度与叶肉导度的比值从0.75左右降至0.36，这个结果表明，通过明确某一植被功能型叶肉导度与气孔导度比值，可以间接确定模型最大叶肉导度的合理取值范围；（3）证实了陆面过程模型添加叶肉导度能改进CO₂施肥效应模拟精度，最大叶肉导度值能影响施肥效应模拟结果，当最大叶肉导度高于0.57 mol m^-2 s^-1后，随最大叶肉导度增加，模拟GPP随大气CO₂浓度增加的增长率呈下降趋势；（4）在月尺度上叶肉导度模拟对最大叶肉导度值的敏感性随不同生长季而不同，在生长盛期的7、8月份最大叶肉导度对叶肉导度模拟结果影响最大，其次是5、6、9月份等生长次盛期，其他月份的影响较小。

Sensitivity analysis of maximum mesophyll conductance parameter in land surface model coupled with mesophyll conductance

Adding mesophyll conductance to the land surface model can effectively improve the accuracy of CO₂ fertilization effect in the model. However, the simulation of mesophyll conductance is affected by the maximum mesophyll conductance parameter and the maximum mesophyll conductance is a key parameter to improve the land surface model for mesophyll conductance and CO₂ fertilization effect simulation. In this study, we added mesophyll conductance to the EALCO (Ecological Assimilation of Land and Climate Observations) model. By changing the maximum mesophyll conductance values artificially, we analyzed the responses of the model outputs to the variable maximum mesophyll conductance values, explored the sensitivity of the maximum mesophyll conductance parameter in the model and compared the results with some existing researches or observation data. Based on the comparisons, we explored the way to optimize the maximum mesophyll conductance parameter of land surface model. The simulation experiment was driven by the oberservation of a typically temperate deciduous broad-leaved forest ecological station in the Harvard Forest Environmental Monitoring site (US-Ha1 site). The results showed that the gross primary production (GPP) simulation accuracy increased with the increase of the maximum mesophyll conductance value. When the maximum mesophyll conductance was greater than 1.0 mol m^-2 s^-1, the improvement of the simulation accuracy was limited, while the GPP simulation accuracy was sensitive when the maximum mesophyll conductance value was less than 1.0 mol m^-2 s^-1. The results also confirmed that there was an obvious linear relationship between mesophyll conductance and stomatal conductance. The changes in the value of the maximum mesophyll conductance affected the slope of linear relationship obviously. When the value of the maximum mesophyll conductance increased from 0.5 mol m^-2 s^-1 to 1.2 mol m^-2 s^-1, the ratio of stomatal conductance to mesophyll conductance decreased from 0.75 to 0.36. The result suggests that by clarifying the ratio of mesophyll conductance to stomatal conductance of plant functional types, the reasonable value range of the maximum mesophyll conductance in land surface models can be determined indirectly. Our results also confirmed that adding mesophyll conductance to the EALCO model could improve the simulation accuracy of CO₂ fertilization effect, and the maximum mesophyll conductance could affect the simulation of CO₂ fertilization effect. When the maximum mesophyll conductance value was greater than 0.57 mol m^-2 s^-1, the growth rate of GPP with the increase of atmospheric CO₂ concentration showed a downward trend with the maximum mesophyll conductance value increasing. We found that the sensitivity of mesophyll conductance simulation to the maximum mesophyll conductance changes was different in different growing seasons. The maximum mesophyll conductance in the peak growth season (July and August) has the greatest impact on the simulation of mesophyll conductance, less impact in the sub-peak growth season (May, June, and September), and little influence in other months.