基于GIS和改进BP神经网络的天然白桦林健康评价

Assessment on Health of Natural Betula platyphylla Forest Based on GIS and BP Neural Networks

结合样地调查及白桦林生态系统相关数据,基于GIS技术平台,利用改进BP神经网络模型,对长白山白河林业局两江林场进行健康评价。利用改进BP神经网络研究复杂的森林生态系统健康状况具有强大优势,运用Leven-berg-Marquardt优化算法比传统的BP神经网络及其它改进算法迭代次数少,收敛速度快,精确度高。评价结果显示:两江整个林场的森林大体处于一般健康状态,占总面积22.43%的白桦林属于非常健康状态,49.93%的白桦林处于一般健康状态,19.06%的白桦林为亚健康状态,而只有8.59%的白桦林是不健康的。其评价结果可为森林可持续经营和多功能利用提供理论支撑。

On the bases of sample investigation and related ecosystem data collected by others,assessment on forest health of Liangjiang Forest Farm in Changbai Mountains was performed by the GIS platform and the improved BP neural network model.Improved BP neural network has great advantages in solving the complex forest ecosystem.Using Levenberg-Marquardt algorithm was effective than the traditional BP and some other improved methods,while it could converge much quickly and has high accuracy.Evaluation results show that the total area of 22.43% of Natural Betula platyphylla forest is very healthy,49.93% of that is healthy,19.06% of that is sub-healthy,while only 8.59% of which is unhealthy.The main purpose of this paper is to provide a solid theoretical support for forest health management and multifunctional use of forests.