*2.2. Data Sources*

Spatial data and non-spatial data were selected for the study, in which spatial remote sensing image data were selected as the primary data for 1964, 1999, 2004, 2014, and 2021. Resolutions were in the order of 2.7 m, 2.7 m, 10 m, 2.5 m, and 2.5 m. After the field survey of the study area, the spatial remote sensing images were interpreted and the land use data were visually interpreted for each year.; According to the Land Use Status Classification (GTB-21010-2017) and the actual situation of the trough valley area, The land use types in

the study area were further divided, where arable land was classified as steep slope, gentle slope, and flat dam (slope > 15◦, 7◦ < slope < 15◦, slope < 7◦); grassland was classified as high cover, medium cover, and low cover, and the others were interpreted according to the images in turn (Table 1). The accuracy of the classified land use types was corrected using ENVI5.0 and verified by combining field research with sampling, and the accuracy of the land use vector map for each period reached 87%, meeting the needs of land use analysis. The non-spatial data were mainly extracted from field research and government statistics.

**Figure 1.** The study area ((**a**) Langxi trough location map; (**b**) Typical settlement distribution map; (**c**) Trough and valley elevation map; (**d**) Slope map of trough and valley).


**Table 1.** Trough land use classification.

#### **3. Methods**

#### *3.1. Settlement Selection and Classification*

The difference in settlement types in the karst valley areas reflects the land-bearing capacity and the human–land relationship in the karst mountains. The classification of settlement types in the valley areas aims to reveal the characteristics of settlement differences in karst valley areas, evolution rules, and driving factors. The classification study helps us grasp the land use changes in different types of settlements. Research on the classification of settlement types usually follows the principles of wholeness, dominance, and the feasibility of development and classifies rural settlements into different types based on the geographical environment, location conditions, economic development, ecological environment, social culture, and farmers' wishes, and then formulates the corresponding optimization strategies [35,49–51]. In this study, based on the avoidance of administrative and large scattered villages, we selected eight typical settlement units in the trough valley region for the study and explored the land use buffer scale changes for individual settlements. Combining the previous research results, in order to grasp the land use evolution pattern guided by human activities in different settlement environments in the trough valley region, considering the geographical differences in the natural environments in which the settlements are located and the types of settlement evolution, drawing on the literature [52–54], we reintegrated the total rate of change and the average annual rate of change with modified formulas and calculated the total rate of change and the average annual net rate of change formulas for the analysis of rural settlement change. Finally, we used the settlement change index for settlement type classification and classified the typical settlement types in the karst trough valley area as follows (Figure 2): expanding settlements (ES: ZengJia, SanCun, and ChuanYan), atrophic settlements (AS: Ganlong and XinCao), disappearing settlements (DS: TaiYangPing), and balancing rural settlements (BS: HeXi and GaoZhai).
