沙棘灌丛林受害程度的预测模型

我国沙棘灌丛林目前的灾害可分为生物灾害和非生物灾害。生物灾害以沙棘木蠹蛾危害为主，是近几年在内蒙古、辽宁、山西、宁夏和陕西等地大面积爆发的一种钻蛀性害虫；非生物灾害主要是立地和气候因素，总体表现为沙棘生长衰退，甚至出现大面积枯死。对辽宁省建平县和内蒙古敖汉旗不同受害程度沙棘灌丛林进行抽样调查，共设50块标准地，分别调查沙棘死亡株率、有叶树冠比例、沙棘木蠹蛾危害株率和20cm根干部沙棘木蠹蛾虫口数等沙棘受害指标，同时调查记录坡向、坡位、坡度、树高、地径和林分密度等多项生跃环境因子。通过模糊综合评判法，对4种受害特征指标进行权重分配，并根据综合评判值的大小，将沙棘灌丛林的受害程度分为健康、轻、中和重4个等级，以坡向、坡位、坡度、树高、地径和林分密度为预测因子，分别构建沙棘灌丛林受害程度BP神经网络模型和Logistic模型，BP神经网络的拟合率达88．1％，预留样本检验准确率为75％，Logistic模型的概率一致率为69．4％。在其它条件相同的情况下，坡位对受害程度的影响是其它因素的4．93倍。

Model of disasters prediction in seabuckthorn shrubberies

The disasters in seabuchkthorn shrubberies, biotic and abiotic disasters. The biotic disasters are mainly caused by seabuckthorn carpenterworms Holcocerus hippophaecolus, which have been a severe boring pest of seabuckthorn in Inner Mongolia Autonomous Region, Liaoning, Shanxi, Ningxia and Shanxi Provinces in recent years. The abiotic disasters include bad site condition and unsuitable meteorological condition. These disasters result in the decline of growing vigor and death eventually in large scale in wide areas in Northern China. An investigation on health status of seabuchkthorn shrubberies was made in Jianping County, Liaoning Province, and Aohan County, Inner Mongolia. Totally fifty sampling plots were surveyed. The field survey includes mortality of seabuckthorn, percentage of crown with leaves, damage rate by the pest, population density of larvae in rootstocks ranging from ground to 20cm underground, average height, bottom diameter of seabuckthorn, direction of slope, position of slope, grades of slope. Based on Fuzzy evaluation, weights of four damage indexes which include mortality of seabuckthorn, percentage of crown with leaves, damage rate by the pest, population density were reset. The degrees of damaged Seabuckthorn shrubberies are measured by healthy, light, medium, severe according to Fuzzy evaluation. With direction of slope, position of slope, grades of slope, height, bottom diameter and population density as predictive factors, developed were BP neural network for classifying and predicting the damage degree of Seabuckthorn shrubberies and Logistic model. The results of prediction show that the qualified rates of fitting and prediction using BP model amount to gg. 1% and 75% respectively. The rate of concordant in Logistic model is 69.4 %. The prediction of BP model agrees with that of Logistic model. Ceteris paribus, the effects on damage degree arising from position of slope is 4.93 times of that arising from other factors.