塔里木河下游河岸柽柳林冠层导度变化特征及模拟

冠层导度(G_c)对植被的蒸腾和光合作用具有重要影响。利用涡度相关仪器实测了塔里木河下游河岸柽柳林地的蒸散发,以及气象因子(温度、湿度、总辐射、光和有效辐射),并利用Penman-Monteith公式计算了柽柳林在2013年生长季的冠层导度。结果显示:柽柳林冠层导度日变化过程在8:00左右迅速增大,于10:30左右达到最大值,之后缓慢下降,18:00左右快速降低;柽柳林冠层导度季节变化过程总体显示,展叶期缓慢上升,落叶期迅速下降,生长盛期缓慢波动下降;研究区,叶面积指数(LAI)是影响柽柳冠层导度季节变化的主要因素,其次为温度(T)、光合有效辐射(PAR)、总辐射(S)、空气饱和差(VPD);四元线性回归方程可以较好地拟合冠层导度与各因子的关系,利用2013年奇数天数据建立回归方程,对偶数天冠层导度值进行模拟和验证,RMSE值为0.169 mm/s,NSE值为0.814,达到了较高的模拟精度。

Variation and predictive simulation of canopy conductance of a Tamarix spp. stand in the lower Tarim River basin

Canopy conductance (G_c) has a crucial influence on vegetation transpiration and photosynthesis. In this study, we measured the evapotranspiration (ET) of a Tamarix spp. stand in the 2013 growing season in the lower Tarim River basin using the eddy covariance technology. Meteorological variables including temperature (T), vapor pressure deficit (VPD), solar radiation (S), and photosynthetically active radiation (PAR) were also measured. We calculated G_c using the inverted Penman-Monteith equation and analyzed its daily and seasonal variation. The relationships between G_c and leaf area index (LAI), T, PAR and VPD were assessed through monadic and multivariate regression analysis. We used data from odd days of the year (DOYs) to build a predictive model and used data for even DOYs to evaluate it. The results showed that:(1) G_c of the Tamarisk spp. stand had a similar variation pattern during the greening-up period (GP), maturity period (MP), and senescence period (SP). G_c increased rapidly in the morning, reached a maximum at 10:30 am, and decreased gradually until 6:00 pm at which point there was a sharp decrease; (2) the seasonal variation of G_c was significant. G_c increased rapidly, decreased slowly, and decreased rapidly with mean values of 0.56 in GP, 1.27 in MP and 0.59 in MP, respectively; (3) LAI was the main driver of G_c, the determinative coefficient was 0.746 in the monadic regression analysis and the partial correlation coefficient reached 0.715, followed by T, PAR, and VPD. Multivariate regression analysis that included all the variables had a determinative coefficient (R²) of 0.79; and finally, (4) the prediction model from odd DOYs had a multiple correlation coefficient of 0.885 with a P value less than 0.01. The model test performed using predicted and calculated G_c over even DOYs resulted in a R² of 0.790, a root mean square error (RMSE) of 0.169 mm/s, and the Nash-Sutcliffe model efficiency coefficient (NSE) of 0.814, indicating that the model had a relatively high accuracy.