Selecting suitable sites for mountain ginseng(Panax ginseng)cultivation by using geographically weighted logistic regression

作     者：HAN Hee JANG Kwang-min CHUNG Joo-sang 

作者机构：Department of Forest EngineeringResources and ManagementCollege of ForestryOregon State UniversityCorvallisOR 97331USA Department of Forest SciencesCollege of Agriculture and Life SciencesSeoul National UniversitySeoul 08826Republic of Korea Research Institute for Agriculture and Life SciencesSeoul National UniversitySeoul 08826Republic of Korea 

出 版 物：《Journal of Mountain Science》 (山地科学学报（英文）)

年 卷 期：2017年第14卷第3期

页      面：492-500页

核心收录：

学科分类：1008[医学-中药学(可授医学、理学学位)] 10[医学] 

基　　金：R&D Program for Forestry Technology funded by Korea Forest Service(Project No.S121012L100100) the framework of international cooperation program funded by National Research Foundation of Korea(2013K2A2A4000649,FY2013) 

主　　题：Panax ginseng Site suitability Logistic regression Geographically weighted logistic regression Geographic Information System South Korea 

摘      要：With the well-being trends to pursue a healthy life, mountain ginseng(Panax ginseng) is rising as one of the most profitable forest products in South Korea. This study was aimed at evaluating a new methodology for identifying suitable sites for mountain ginseng cultivation in the country. Forest vegetation data were collected from 46 sites and the spatial distribution of all sites was analyzed using GIS data for topographic position, landform, solar radiation, and topographic wetness. The physical and chemical properties of the soil samples, including moisture content, p H, organic matter, total nitrogen, exchangeable cations, available phosphorous, and soil texture, were analyzed. The cultivation suitability at each site was assessed based on the environmental conditions using logistic regression(LR) and geographically weighted logistic regression(GWLR) and the results of both methods were compared. The results show that the areas with northern aspect and higher levels of solar radiation, moisture content, total nitrogen, and sand ratio are more likely to be identified as suitable sites for ginseng cultivation. In contrast to the LR, the spatial modeling with the GWLR results in an increase in the model fitness and indicates that a significant portion of spatialautocorrelation in the data decreases. A higher value of the area under the receiver operating characteristic(ROC) curve presents a better prediction accuracy of site suitability by the GWLR. The geographically weighted coefficient estimates of the model are nonstationary, and reveal that different site suitability is associated with the geographical location of the forest stands. The GWLR increases the accuracy of selecting suitable sites by considering the geographical variations in the characteristics of the cultivation sites.