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

Introduction The process of climate regionalization divides an area into smaller regions that are homoge- neous with respect to a specified climatic characteristic. The Ko¨ppen-Geiger climate classifica- tion [1, 2], which broadly classifies regions by monthly air temperature and precipitation, is one of the most widely used climatic classification systems in the world. It is a global and rela- tively coarse climate classification scheme. However, the climatic regionalization of Qinling- Daba Mountains (QDM) is extremely challenging due to complex topographical conditions and large mountains. According to the Koppen classification system [3], eight basic climate types are mixed in this region. In fact, QDM has served as the natural geographical boundary of north/south China and the climate boundary of subtropical/temperate zones from a very early time [4, 5]. The accurate division of climatic boundaries was significant in both general and applied climatology [6]. By comprehensive analyzing each element of the environment, the spatial stratified heterogeneity of natural surface in QDM be correctly reflected. In China, this domain has been explored for more 80 years, resulting in the accumulation of numerous results [7–10]; hundreds of scholars have proposed and developed many subtrop- ical/temperate zone boundary division methods based on climate, vegetation, and soil [3, 11– 14]. The location of the subtropical/temperate zone boundary represents an essential issue of Chinese physical geography [10]. Due to the differences in the delimitation indicators and

research purposes, the specific location of this climate boundary remains controversial [15]. Generally, there are three main opinions on the location of the climate boundary in QDM: the north slope of Qinling Mountains, the main ridge of Qinling Mountains (the watershed of the Yellow River and the Yangtze River), and the southern slope of the Qinling Mountains at an altitude of 800-1000 m [16]. Zhu Kezhen (1958) proposed that the Qinling-Huaihe river area was the subtropical/temperate boundary [17]. He considered temperature as the key indicator along with temperature-related indicators, including the accumulated temperature, the coldest month average air temperature, etc. Chen Xianji (1982) also delimited the subtropical/temper- ate boundary in the Qinling-Huaihe river, based on temperature-related indicators. However, Qiu Baojian, Zheng Du, and others suggested that the boundary of the subtropical/temperate zone should be calibrated on the main ridge [18, 19]. In addition to temperature-related indi- cators, vegetation and soil also serve as auxiliary indexes. However, others hold the opinion that the boundary was mainly located on the southern slope of the Qinling Mountains. Zhang Jinquan suggested that vegetation and soil should be mainly considered because these parame- ters could better reflect the long-term development of nature. The boundary was mainly located on the southern slope of the Qinling Mountains at an altitude of 800 m [10]. Although much of the controversy is caused by differences in the selection of index systems

and the research method of climatic regionalization schemes [20], the most important reason is the complexity, transition, and heterogeneity of QDM. In fact, the process of change from the subtropical zone to temperate zone is gradual, and the boundary of the subtropical/temper- ate zone cannot be drawn as a line [21, 22]. Scholars were often misled by the term “boundary line” and ignored the properties of the transition zone [23]. The concept of the subtropical/ temperate zone boundary should be understood as a wide or narrow transition zone [24]. This is also the reason why different scholars delimited the boundary with significant differences. Therefore, to study and analyze the characteristics of the QDM, focusing on the transition zone rather than the exact location of the boundary between subtropical and temperate zones would be more scientific. To identify the location of the transition zone of the QDM, a new workflow for comprehensive physical regionalization was proposed in this paper. Methodologically, with the development of machine learning, an increasing number of sta- tistical models used in many studies to define climate regions in the world. These models were PLOS ONE Climate regionalization based on Geodetector-SVM model PLOS ONE | https://doi.org/10.1371/journal.pone.0241047 November 2, 2020 2 / 16