Exploring on the climate regionalization of Qinling-Daba mountains based on Geodetector-SVM model

subtropical and temperate vegetations, such as Quercus wutaishanica forest and Quercus fabri forest). The higher the value of q-statistics, the stronger impact of a single factor on the spatial distribution of vegetation. According to the previous regionalization scheme, the average q- statistic value is approximately 0.33 between climatic zones [36]. In this study, the variables with q-statistic value greater than 0.3 were selected. Moreover, only four variables in group II (Pinus tabuliformis forest and Pinus massoniana forest) had q-statistic values greater than 0.3. To assure the equality of data for each training session, the first four variables with q-statistic values greater than 0.3 were selected. The third step was to build a comprehensive physical regionalization model based on the SVM to identify the boundary of the subtropical zone and temperate zone in QDM. Specifically, the training sets (including index system) and the labels of training sets (the vegetation types) were used as input in the SVM model, and the zoning maps were output from the model. The experiment was conducted for every paired vegetation, and the subtropical northern boundaries were extracted from each output of the SVM model. The boundary refers to the line between the two partitions of the output of SVM model. The last step was to validate the results using plant community survey data, comprehensively ana- lyze the results, and delimit the location of the transitional zone. Geodetector. The Geodetector model proposed by Wang et al. is a novel and suitable spa-

tial analysis method to measure the degree of spatial stratified heterogeneity [34, 35], which has been widely applied in many fields in recent years [45–47]. In this paper, it was applied to Fig 3. Workflow of the identification of transition zone. https://doi.org/10.1371/journal.pone.0241047.g003 PLOS ONE Climate regionalization based on Geodetector-SVM model PLOS ONE | https://doi.org/10.1371/journal.pone.0241047 November 2, 2020 7 / 16