玉米蒸散发变化的归因——基于贝叶斯原理参数优化的双源模型

Attribution for Evapotranspiration Changes in Maize:A Two-source Model Based on Bayesian Principle Parameter Optimization

蒸散发(ET)变化的归因分析对理解蒸散发的控制机理具有重要意义,是建立可靠的蒸散发模型和预测蒸散发变化的基础。通过贝叶斯方法,用差分进化自适应都市(DREAM)算法对双源蒸散发模型(SW)参数优化,使其模拟值与在玉米农田中获得半小时ET观测值达到最佳模拟效果。然后将涡度相关法与SW模型相结合,利用泰勒偏微分方程定量评估了黑河流域大满站玉米区域气候和植被变量对土壤蒸发(ET_(s))和植物散发(ET_(c))的相对贡献。与2016年相比,2017和2018年平均ET_(s)分别增加了2.56 W/m~2和72.90 W/m~2,2017年ET_(c)增加了24.41 W/m~2,2018年减少了13.38 W/m~2。2017年和2018年土壤表面可用能量(A_(s))对ET_(s)变化的贡献分别为124%和95.4%,冠层阻力(r_(s)^(c))对ET_(c)变化的贡献分别为68.8%和71.9%。参数优化后的SW模型与实测值拟合良好,能够反映ET的动态变化。ET的年际变化受气候和植物的复杂相互作用的影响,A_(s)和饱和水汽压差是驱动ET_(s)年际变化的主要因子,而冠层上方可用能量和r_(s)^(c)是驱动ET_(c)年际变化的主要因子。

Attribution analysis of changes in evapotranspiration (ET) is important for understanding the mechanisms that control ET,which is necessary for reliable modeling and predicting.Based on the Bayesian method,the parameters of the two-source evapotranspiration model(SW) were optimized by the Differential Evolution Adaptive Metropolis (DREAM) algorithm to achieve the best simulation effect between the simulated value and the half hour ET observed in maize field.Then,the relative contributions of regional climate and vegetation variables to soil evaporation (ET_(s)) and plant transpiration (ET_(c)) at Daman Station in Heihe River Basin were quantitatively evaluated by using the Eddy covariance method and the SW model.Compared with the data in 2016,ET_(s)increased by an average of 2.56 W/m~2and 72.90 W/m~2in 2017and 2018,respectively.ET_(c) increased by 24.41 w/m~2in 2017 and decreased by 13.38 w/m~2in 2018.In 2017 and 2018,the contribution of soil surface available energy (A_(s)) to ET_(s) variation was 124%and 95.4%,and the contribution of canopy resistance (r_(s)^(c)) to ET_(c) variation was 68.8%and 71.9%,respectively.The SW model after parameter optimization fits well with the measured values and can reflect the dynamic changes of ET.The interannual changes in ET were influenced by complex interactions between climate and vegetation variables.A_(s ) and saturated vapor pressure deficit were the main factors driving the interannual variation of ET_(s),and the available energy above the canopy and r_(s)^(c) were the main factors driving the interannual variation of ET_(c).