常绿林“导度—光合”模型斜率参数优化与蒸腾估算改进

Optimization of the Stomatal Conductance Slope in the Conductance-Photosynthesis Model and Improved Estimation of Transpiration in Evergreen Forests

“导度—光合”模型广泛应用于植被蒸腾估算,其中气孔导度斜率是模型的核心参数,通常用一个特定于生物群系的固定值进行参数化。然而,有研究指出气孔导度斜率存在季节性变化,故常数化气孔导度斜率的方案将导致植被蒸腾估算产生较大的不确定性。因此,如何优化气孔导度斜率的参数方案是提升植被蒸腾估算精度的关键。已有研究表明,使用叶面积指数对气孔导度斜率进行动态参数化可以有效改进对落叶林植被蒸腾的估算,但目前尚不清楚该方法是否同样适用于林冠季节变化不明显的常绿林。此外,最优性原理进一步解释气孔导度斜率为温度的函数。因此,利用温度模拟气孔导度斜率的效果是否优于叶面积指数也有待研究。针对以上问题,选取6个FLUXNET常绿林站点研究气孔导度斜率与叶面积指数和气温的关系,并对比2种参数化方案的模拟结果。结果显示,气孔导度斜率在各站点都随叶面积指数和温度的变化而变化,二者均呈现显著的负相关关系,动态参数化方案结果均优于常量气孔导度斜率静态参数化方案。这表明在常绿林中利用叶面积指数对气孔导度斜率模拟仍然有效,但温度对气孔导度斜率的解释能力(R^(2)=0.45±0.12)比叶面积指数的(R^(2)=0.28±0.23)更强。进一步对比基于叶面积指数和温度参数方案所得的气孔导度和植被蒸腾,发现在日尺度上基于温度的方案比基于叶面积指数的结果改进更明显,显著降低了植被蒸腾估算的均方根误差(26.0%±24.4%)。以上结果说明基于温度的气孔导度斜率参数化方案具有可行性,并有助于从机理层面改进常绿林生态系统的气孔导度和植被蒸腾动态的模拟。

The conductance-photosynthesis(gs-A)model is widely used for estimating the transpiration rate(ET_(c)).The stomatal conductance slope(g_(1))in the gs-A model is crucial,and is usually parameterized with a PFTspecific g_(1).However,because there is seasonal variation in g_(1),the scheme of constantizing g_(1)results in large potential uncertainties in ET_(c)estimation.Therefore,optimizing the parameterized scheme of g_(1)is key for improving the estimation of ET_(c).Previous studies have shown that the dynamic parameterization of g_(1)using remote sensing-based Leaf Area Index(LAI)can effectively improve the accuracy of the estimated ET_(c)in deciduous forests.However,it is still not clear whether this method is applicable to evergreen forests,in which the canopy shows slight seasonal variability.In addition,the First-Principles Theory indicates that g_(1)can be expressed as a function of temperature.Hence,whether temperature can be used to simulate g_(1)with a better performance than that when using LAI requires further investigations.In view of the above questions,six FLUXNET evergreen forest sites were selected to investigate the relationships between g_(1),and LAI and air temperature,and the ET_(c)simulation results under the two parameterization schemes were compared.Results showed that g_(1)varied with both LAI and temperature at each site,showing a significant negative correlation,while the results of the dynamic parameterization scheme(DYN)were better than those of the constant g_(1)scheme(FIX).This showed that the simulation of g_(1)using LAI is still effective in evergreen forests,but the explanation of temperature on g_(1)(R^(2)=0.45±0.12)is stronger than that of LAI(R^(2)=0.28±0.23).Further comparing the gs and ET_(c)obtained using the two schemes,we found that the scheme based on temperature performed better than that of LAI at the daily scale,with the Root Mean Square Error(RMSE)of the ET_(c)estimation being significantly reduced(26.0%±24.4%).This study emphasizes the feasibility of the temperature-based parameterized scheme of g_(1)and helps to fundamentally improve the simulation of gs and ET_(c)in evergreen forest landscapes.