Abstract:Based on the data gathered of the growth of E.grandis×E.urophylla stands and of the soil under the stands and with the aid of a new method-artificial neural network,study was conducted on leading soil factors inf luencing growth and yield of E.grandis×E.urophylla in South Fujian hilly area.The BP model was established simulating the relationship between the leading soil factors and the growth of E.grandis×E.urophylla.The results demonstrate that the leading soil factors,i.e.,the soil capillary moisture-holding capacity,humus layer thickness,soil layer thickness and available phosphorus,are in a nonlinear mapping relationship with the growth of E.grandis×E.urophylla.The accuracy of BP model in simulating the fast growth and high yield of E.grandis×E.urophylla in South Fujian hil y area is 96.39%,fairly higher than the traditional linear regression model.Therefore,this study not only provides a basis for improving E.grandis×E.urophylla,but also opens up a new train of thought in the application of artificial neural network to the research of selecting forest-suitable sites.