潇河冬小麦水肥生产函数偏最小二乘回归建模及分析

Modeling and Analysis by Partial LeastSquares for Water-fertilizer Production Function of Winter Wheat in Xiaohe Irrigation District

在最小二乘回归失效的情况下,将偏最小二乘回归应用于山西潇河灌溉试验站冬小麦水肥生产函数建模。该方法可以克服多重相关引起的危害,建立合理的、具有较强的稳定性和预测能力的水、氮、磷生产函数模型。由模型分析可知,水、氮、磷三因素都具有明显的增产效应。水肥耦合效应表现为水肥对产量的协同作用和氮、磷之间的竞争关系。随着作物耗水量的增加,氮、磷的利用效率得到提高;而增加氮、磷投入亦能提高水分利用效率。此外,还提出了氮、磷有效组合区域的概念。

Partial least-squares regression is applied in modeling of fertilizer-water production function of winter wheat in Xiaohe irrigation station, Shanxi province when least-squares regression fails and then the water nitrogen phosphor production model of 2002～2003 is acquired. The simulation result justifies that partial least-squares regression can solve the problems caused by multicollineari ty in multi-regression and the model established is reasonable. According to it, the production-increasing effect of evapotranspiration, nitrogen and phosphor is remarkable. The coupling effect between water and fertilizer manifests that they can boost the production increasing effect mutually. Water use efficiency will increase when the nitrogen and phosphor are increased and vice versa. Fertilizer use efficiency will decrease when nitrogen and phosphor are increased. And nitrogen and phosphor are competitive to some degree. In addition, a so called conception ＂effective compounding zone of nitrogen and phosphor＂ is presented in this paper.