中文摘要：

利用波文比能量平衡法对天山北麓绿洲荒漠过渡带芨芨草地的夏季蒸散量进行测定，采用通径分析与相关分析相结合的方法探讨了蒸散与主要环境因子之间的关系，建立了适合该地区的蒸散预测模型。结果表明：（1）各种环境因子对蒸散影响的大小排序为：净辐射（Rn）〉土壤热通量（G）〉空气温度（T气）〉空气相对湿度（RH）〉5cm土壤温度（T±-5cm）〉实际水汽压（ea）〉风速（Vwind）〉5cm土壤含水率（S±-5cm）；（2）各因子通过Rn对蒸散产生的间接作用都大于其自身直接作用，反映出净辐射是制约蒸散大小的主导因子，是决定干旱区蒸散量的关键；（3）土壤热通量对蒸散的直接负效应远小于通过其他因子的间接正效应，出现其直接负效应与综合效应相反的结果；（4）建立并经过检验的蒸散预测模型表明：与温暖湿润区蒸散依靠水、热并重情况不同的是，极端干旱区主要依靠热量因子强度的增加，才能有较大的蒸散量。

英文摘要：

The Achnatherum splendens grassland evapotranspiration of oasis-desert ecotone in the northern Tianshan Mountain was calculated using Bowen ratio equilibrium energy method, then the influence of different weather conditions on the relationship between evapotranspiration and environmental factors were analysed, based on the ob- served meteorological data. Furthermore, evapotranspiration forecasting models were established. The results showed as follows： （1）the significance of impact of environmental factors on evapotranspiration implied that： net radiatio （ Rn ） 〉 soil heat flux （ G ） 〉 air temperature〉 air relative humidity （ RH ） 〉 5 cm soil temperature 〉 actual vapor pressure （ ea ） 〉 wind speed （ Vwind ） 〉 5 cm soil moisture. Therefore, net radiation （ Rn ）, soil heat flux （ G ）, air temperature, and relative humidity （ RH ） were the four major factors, whose absolute values of the correlation coefficient were above 0.44. However, except for 5 cm soil moisture （Ssoil-5cm ）, all the correlation between each climate factor and evapotranspiration reachd a very significant level. （2）The indirect effect of each factor produced by Rn on evapotranspiration was greater than its direct effect, meaning that net radiation is the principal factor determining the scale of evapotranspiration. Especially, the absolute value of indirect effect coefficient of G, air temperature and RH produced by Rn were between 0.4 and 0.74, which were much larger than their own direct coefficients. （3） The direct negative effect of G on evapotranspiration was much smaller than other indirect positive effects, causing its comprehensive effect to be negative. For instance, the indirect effect coefficient of G produced by Rn was up to 0.744 1, which was 3.34 times as big as the direct negative effect, leading to its comprehensive effect being positive. （4） The evapotranspiration forecasting models in which evapotranspiration was chosen as dependent variable and each environme