Abstract: Low rainfall, fragile ecology, and economic poverty are typical characteristics of the Loess Plateau. Jia county of Yulin City, Shaanxi Province, is a typical county in the semi-arid Loess Plateau, and is the main battle? field for ecological restoration and poverty alleviation in China. Following the theory of human-land system vulnera? bility, using TOPSIS model, this studymeasured the vulnerability value of microscale rural human settlement and natural, dwelling, human, support and social subsystems during the four historical periods from 1980 to 2017 (1980s, 2000, 2008, 2017) based on 65 sample villages, 451 household questionnaires, and 65 thematic question? naires for key village figures. Additionally, the spatial and temporal differentiation characteristics of vulnerability of rural human settlements in the research area, based on GIS spatial analysis technology, were also discussed,.The study results show the following.(1)The average vulnerability of rural human settlements in the 65 sample villages decreased from 0.66 to 0.45, and the vulnerabilities of dwellings, and natural and social subsystems were significant? ly reduced, with the mean vulnerability values for dwellings and natural and social subsystems in the 65 sample vil? lages falling to 0.38, 0.42, and 0.47, respectively.However, the mean value of support and human system vulnerabili? ty in the 65 sample villages remained unchanged or rose above 0.54. (2)Based on the 65 sample villages, the spatial differences of vulnerability of the comprehensive system and the five subsystems of rural human settlements all showed expanding trends with different ranges. Among these, spatial differences of the integrated system and the dwelling subsystem increased by nearly 12 percentage points to 15.31% and 17.88%, respectively. The natural sub? system (30.77%) had the largest spatial difference, followed by the support subsystem (21.61%), both of which in? creased by a factor of two compared to the 1980s. The spatial differences in vulnerability of both human and social subsystems increased slightly, rising by only 3 percentage points since the 1980s. (3)The comprehensive system of rural human settlements has evolved from a uniform distribution of severely vulnerable sample villages to a main dis? tribution of sample villages with medium- and low-vulnerability, and 24.62% of the highly vulnerable sample villag? es existed. Dwelling, natural, and social subsystems all presented similar evolution processes, shifting from a global coverage of high-vulnerability sample villages to a widespread distribution of low-vulnerability sample villages, with only 1.54% of the high- vulnerability villages left in the dwelling subsystem. In contrast, the human subsystem evolved from a main distribution of medium- and low-vulnerability sample villages to a global coverage of high-vul? nerability sample villages. Additionally, thespatial pattern of the support system is unstable,with the low vulnerabili? ty level concentrated in sample villages located along major traffic lines or in the administrative centers of towns. At present, the proportion of high-vulnerabilitysample villages of human and supporting subsystems have both been ex? ceeded two-thirds. Based on the above findings, this study discusses the temporal and spatial process of rural hu? man settlement vulnerability from the micro scale, which better reflects the development and unbalanced character? istics of rural human settlements. Therefore, this paper suggests that future work should not only focus on the vulner? ability of the human and support subsystems, but also on spatial imbalances of the natural, support, and dwelling systems.

Key words: rural human settlement, vulnerability, spatial and temporal pattern, TOPSIS model, Jiaxian county