基于SOM神经网络的五指山市森林健康评价

Forest health assessment based on self-organizing map neural network in Wuzhishan City

SOM神经网络具有无样本自组织学习的功能,能从输入数据中寻找规律,使结果客观、真实。本研究采用1993-2008年五指山市四期一类调查数据,运用SOM神经网络分析森林动态变化,对五指山市森林健康进行评价,并探讨了森林健康等级与龄组和郁闭度之间的关系。结果表明:1993年五指山市森林健康水平较低,Ⅰ等质量占据绝对主导地位,随后森林质量逐步提高,至2008年,五指山市森林健康等级,Ⅲ、Ⅳ等占据主体,森林健康水平得到很大提高,同时在一定范围内,森林表现出小班龄组越大,郁闭度越大,健康等级越高。以上评价结果可为五指山市的森林可持续经营利用提供理论支撑。

SOM neural network can organize learning without any sample, and it also can find the patterns from the input data and make the result objective and true. The research uses SOM neural network to analysedynamic change of forest in Wuzhishan City with the survey data from 1993-2008 years. And discuss the relationship among forest health level; age group and crown density. The result shows that, in 1993 the forest health level of Wuzhishan City is low, and the first level occupied the leading place, then to 2008, the forest health level is increased. It shows that the third and fourth level Occupied the main position. And in a reasonable change range, forest health level is increased with age and crown density. These results can be toke to help forest management in Whuzhishan City.