干旱区生态环境知识资源中心(Arid): 天然草地分类的Bayes判别分析法

Arid

	天然草地分类的Bayes判别分析法
其他题名	Bayes Discriminant Analysis Method of Natural Grassland Classification
	文畅平; 白银涌; 曾娟娟; 苏伟
来源期刊	中国草地学报
ISSN	1673-5021
出版年	2016
卷号	38 期号:3 页码:50-55
中文摘要	基于多元统计分析理论,综合考虑天然草地的生态状况、可再生产能力和经济条件等因素,选择植被覆盖度、可食风干牧草产量、牧草利用率、草地可利用面积系数、草群中优良牧草比率和羊单位需草地面积6个参数作为判别因子,建立了天然草地分类的Bayes判别分析法。将天然草地分为温性荒漠类、温性草原类和低山地草甸类3个类别,并作为Bayes判别分析的3个正态总体,以新疆准噶尔盆地西部地区的31个天然草地为样本(其中18个为训练样本),以Bayes线性函数值和后验概率对样本所归属的总体进行识别,将建立的模型对训练样本进行回判,以回代误判率对模型进行检验。研究表明,Bayes判别分析模型可以很好地反映不同类型天然草地间的资源属性,对训练样本的回代误判率为0;对另外13个天然草地样本的分类结果与灰色关联度分析法、投影寻踪法、投影寻踪动态聚类法、集对分析法、SOFM神经网络法以及突变级数法等的分类结果基本一致。
英文摘要	The ecological situations, reproduction abilities and economic conditions of natural grassland were taken into account synthetically, and six parameters including vegetation coverage, dry edible forage grass production, forage grass utilization rate, available grassland area coefficient, forage grass utilization rate, grassland area for every sheep unit were selected as discrimination factors. Bayes discrimination analysis method for natural grassland classification was put forward based on multivariate statistical analysis theory. The natural grassland was divided into 3 grades, i.e. temperate desert type, temperate steppe type and lowland mountain meadow grassland type, and regarded as three normal populations for Bayes discrimination analysis; 18 natural grassland samples from western Junggar Basin, Xinjiang, China, were selected as the training samples to present Bayes linear discrimination functions and posterior probability function, and the Bayes function values and posterior probability of samples were used to recognize which population the sample belongs to. The established Bayes discrimination analysis model was used to back-discriminate the training samples, and verified by the ratio of mistake-discrimination. The study indicated that the Bayes discrimination analysis model can better reflect the grassland resources attributes between natural grassland samples, and the ratio of mistake-discrimination of training samples was zero. The classification results of the additional 13 natural grassland samples agreed well with that of grey relational analysis method, projection pursuit method, projection pursuit dynamic cluster method, set pair analysis method, SOFM neural network method, and catastrophe progression method.
中文关键词	天然草地 ; 分类 ; Bayes判别分析法
英文关键词	Natural grassland Classification Bayes discrimination analysis method
语种	中文
国家	中国
收录类别	CSCD
WOS类目	AGRICULTURE MULTIDISCIPLINARY
WOS研究方向	Agriculture
CSCD记录号	CSCD:5722190
资源类型	期刊论文
条目标识符	http://119.78.100.177/qdio/handle/2XILL650/235323
作者单位	中南林业科技大学土木工程与力学学院, 长沙, 湖南 410004, 中国
推荐引用方式 GB/T 7714	文畅平,白银涌,曾娟娟,等. 天然草地分类的Bayes判别分析法[J],2016,38(3):50-55.
APA	文畅平,白银涌,曾娟娟,&苏伟.(2016).天然草地分类的Bayes判别分析法.中国草地学报,38(3),50-55.
MLA	文畅平,et al."天然草地分类的Bayes判别分析法".中国草地学报 38.3(2016):50-55.