气候变化领域集成服务门户(CCPortal): Reconstructing the historical patterns of forest stand based on CA-AdaBoost-ANN model

CCPortal

DOI	10.1016/j.foreco.2020.118518
	Reconstructing the historical patterns of forest stand based on CA-AdaBoost-ANN model
	Zhan X.; Yu S.
发表日期	2020
ISSN	0378-1127
卷号	478
英文摘要	Understanding the long-term interaction between humans and the environment requires the reconstruction of historical forest dynamics. However, the accuracy of methods to reconstruct the patterns of historical forest stands using resurvey data has not been comprehensively evaluated. Previous simulations and statistical assessments of forests have focused on reconstructing land cover or single layer, but are not congruent with the specific data typically obtained by field survey. Additionally, most reconstruction studies have not reconstructed the patterns of the forest stand or the tree species distribution. In this work, we applied cellular automata (CA), which integrated artificial neural networks (ANN), to describe tree species dynamics in response to forest changes in Heishiding Nature Reserve of Guangdong province in south China. The top 70 dominant tree species represented by importance value (IV) and actual distribution were considered. The relationship of the current and historical forest distributions was modeled using an AdaBoost-based back-propagation neural network algorithm. The area under the receiver operating characteristics curve (AUC) values for the 70 studied tree species ranged from 0.8556 to 0.9981, which suggested that the reconstructed patterns were reasonable. The simulation of the forest stands of a 50-ha plot in 2012 achieved an overall accuracy of 94.14% and a Lee Sallee index of 0.7974. The results indicated that this method can be further improved to obtain a more accurate pattern of tree species by adjusting the CA transition rules and parameters. Overall, the results demonstrate that the CA-AdaBoost-ANN model can be applied as an effective path to obtain a distribution map of historical forest types, which will be extremely useful for effective forest management. © 2020 Elsevier B.V.
英文关键词	Artificial neural networks; Cellular automata; Historical reconstruction; Spatiotemporal modeling; Vegetation dynamics
语种	英语
scopus关键词	Adaptive boosting; Backpropagation; Cellular automata; Cellular neural networks; Population distribution; Back propagation neural networks; Distribution maps; Forest distribution; Guangdong Province; Long-term interaction; Overall accuracies; Receiver operating characteristics; Statistical assessment; Forestry; artificial neural network; back propagation; cellular automaton; forest ecosystem; forest management; land cover; nature reserve; reconstruction; Accuracy; Construction; Data; Distribution; Dynamics; Forestry; Patterns; Trees; China; Guangdong; Heishiding Nature Reserve
来源期刊	Forest Ecology and Management
文献类型	期刊论文
条目标识符	http://gcip.llas.ac.cn/handle/2XKMVOVA/154960
作者单位	Department of Ecology, School of Life Sciences/State Key Laboratory of Biocontrol, Sun Yat-sen University, Guangzhou, 510275, China
推荐引用方式 GB/T 7714	Zhan X.,Yu S.. Reconstructing the historical patterns of forest stand based on CA-AdaBoost-ANN model[J],2020,478.
APA	Zhan X.,&Yu S..(2020).Reconstructing the historical patterns of forest stand based on CA-AdaBoost-ANN model.Forest Ecology and Management,478.
MLA	Zhan X.,et al."Reconstructing the historical patterns of forest stand based on CA-AdaBoost-ANN model".Forest Ecology and Management 478(2020).