基于改进型B-P神经网络的西天山云杉林生物量估算  

Picea Schrenkiana Forest Biomass Estimate in the West Tianshan Mountain Based on Improved B-P Neural Network

作　　者：袁野[1,2] 李虎[1] 刘玉峰[1] 

机构地区：[1]福建师范大学地理科学学院,福建福州350108 [2]兰州大学资源环境学院,甘肃兰州730000

出　　处：《福建师范大学学报（自然科学版）》2011年第2期124-132,共9页Journal of Fujian Normal University：Natural Science Edition

基　　金：国家基础科学人才培养基金资助(J0830521)

摘　　要：以西天山云杉林为例,在实地调查的基础上,以NDVI等七个植被指数和TM/ETM数据主成分变换后的前5个主成分为自变量,应用B-P人工神经网络技术,建立了西天山尼勒克林区遥感影像数据与云杉林生物量实地数据之间的神经网络模型.通过训练和仿真,与实地数据进行比较,结果表明,云杉林生物量估算值与实际值之间的平均相对误差为8.21%,精度较高.It takes Picea schrenkiana as an example.With the application of B-P artificial neural network technology,using seven kinds of vegetation index such as NDVI and the former five PCAs that come from TM/ETM principal component transform,a neural network model was established based on the data of remote sensing and field measurements in Nileke west of the Tianshan Mountains.After training and simulation,compared this model with the field measurements,the result shows that its average relative error between estimating value and actual value is 8.21%,which proved to be higher accuracy.

关 键 词：B-P神经网络 天山云杉林 生物量 遥感植被指数 

分 类 号：S718.55[农业科学—林学]