基于SOM神经网络的通辽市库伦旗森林健康评价

Forest Health Assessment in Hure Banner of Tongliao City Based on SOM

森林健康是森林可持续经营的前提。在森林健康众多研究方法中,SOM神经网络(Self Organizing Maps)能够减少主观因素的影响,是较为先进的人工智能评价方法。文章对内蒙古自治区通辽市库伦旗森林健康进行评价,运用主成分分析法选取评价指标,建立森林健康评价指标体系,运用SOM神经网络模型进行聚类分析,结论如下:(1)优质小班、健康小班面积之和占比53.72%,库伦旗乔木林总体健康程度较好;(2)以起源划分,人工林中优质和健康小班面积占比51.89%,天然林中96.72%都是优质小班;(3)以龄组划分,幼龄林和中龄林小班健康情况好于其他小班;(4)以优势树种划分,榆树(Ulmus pumila L)和樟子松(Pinus sylvestris var. mongolica Litv.)小班健康情况好于杨树(Populus L.)、油松(Pinus tabuliformis Carr.)。

Forest health is the premise of sustainable forest management. Among many research methods of forest health, SOM(self organizing maps) can reduce the influence of subjective factors and is a more advanced artificial intelligence evaluation method. In this paper, the forest health of Hure Banner in Tongliao City of Inner Mongolia was evaluated. The principal component analysis method was used to select the evaluation index, and the forest health evaluation index system was established. SOM was used for cluster analysis. The conclusions are as follows:(1) The sum of high-quality subcompartments and healthy subcompartments accounts for 53.72 %, and the overall health of arbor forest in Hure Banner is good;(2) According to the origin, the area of high-quality and healthy subcompartments in plantations accounts for 51.89 %, and 96.72 % in natural forests are high-quality subcompartments;(3) Divided by age groups, the health status of young forest and middle-aged forest subcompartments is better than other subcompartments;(4) In terms of dominant tree species, the health status of Ulmus pumila L and Pinus sylvestris var. mongolica Litv.was better than that of Populus L. and Pinus tabulaeformis Carr.