摘要
森林生物量的估测是全球变化研究的基础,而遥感宏观、综合、动态、快速的特点决定了基于遥感的生物量模型为森林生物量估测的发展方向,目前的遥感生物量估测方法大多基于回归分析,需要预先假设、事后检验,仅为经验性的统计模型。神经网络的分布并行处理、非线性映射、自适应学习和容错等特性,使其具有独特的信息处理和计算能力,在机制尚不清楚的高维非线性系统体现出强大优势,可以用于遥感生物量估测。文章在野外调查的基础上,尝试应用BP网络和RBF网络技术,建立广州TM遥感影像数据与森林样方生物量实测数据之间的神经网络模型,通过训练和仿真,与生物量实测数据进行比较。结果表明,在独立样地估测中,人工神经网络估测的相对误差均小于15.18%,获得了满意的效果。而RBF网络与BP网络相比,在识别精度上、稳定性、速度上,均优于BP网络,其最大相对误差不超过10.12%,平均相对误差为4.76%。可见应用神经网络方法的“黑箱”操作虽然难以归纳出指导性规律,但可以获得很高的精度。尤其RBF网络,在训练完成后,可以应用该模型进行大区域生物量估算,对于森林的规划及管理具有深远意义。
In this paper, biomass value was obtained from 80 plots （30 m×30 m）, including 40plots for broadleaved forests and 40 plots for coniferous forests, respectively. All of plots were positioned accurately using Global Positioning System, the Landsat TM image for Guangzhou area was acquired on Mar 9, 2004. The satellite image was revised. Using the technology of artificial neural network, two different network models were established based on the data of Guangzhou remote sensing and the field measurements, for estimation of the forest biomass, the results shows that the relative errors of estimate value all were less than 15.18%. Comparing with BP network, the RBF has more advantages in application： Its maximal relative error less than 10.12%, its mean relative error was 4.76%, which proved to have a much higher accuracy.
出处
《生态环境》
CSCD
北大核心
2007年第1期108-111,共4页
Ecology and Environmnet
基金
国家教育部"985工程"GIS与遥感的地学应用科技创新平台项目(105203200400006)
