基于径向基函数的神经网络对森林碳空间分布的模拟

Modeling the Spatial Distribution of Forest Carbon Storage by Neural Network Based on Radial Basis Function

利用径向基神经网络,结合森林资源清查的930个样地调查数据和对应的TM影像数据,选取与森林生物量相关性较大的3个植被指数TM4/57、ARVI和KT2作为神经网络的输入变量,对临安市森林碳储量的空间分布进行模拟。结果显示,利用径向基神经网络较好地重建了森林碳储量空间分布和变化,模拟结果与样地实测值间的一致性好,为区域森林碳储量的估测研究提供了方法支持。

By means of applying the radial-basis-function-neural-network（RBFnn） method and integrated with the field survey data from 930 sample plots obtained by the forest inventory and the corresponding TM images,the spatial distribution of the forest carbon storage of Lin′an Municipality was simulated by taking 3 three vegetation indices,i.e.,TM4/57,ARVI and KT2 as the input variables.The results showed the radial-basis-function-neural-network（RBFnn） method could accurately generate the spatial distribution and the variation of forest carbon storage,and there was a very good consistency between the simulated results and the data obtained from the field survey,which provided the predictive studies of forest carbon storage with methodological reference.