Population size change is an important driving force in the evolution of land use and desertification in arid and semiarid areas. The relationship between population size change and land-use change was analyzed using a multi-scale system at the prefecture and township levels. The indirect effect of population size change on PM10 emissions from sandstorms was also discussed. The results were as follows: (1) From 2000 to 2024, population density in the Tarim Basin continued to increase and became increasingly concentrated, forming a spatial pattern of “tower elevation, desert belt formation, and corps jump.” Population growth was highly coordinated with the expansion of cultivated land and construction land and the contraction of bare land at the regional scale. (2) The effect of population changes on land use showed significant spatial non-stationarity, with a pattern of “belt around desert, different from east to west, and heterogeneous from north to south.” The correlation intensity gradually increased from the desert hinterland to the marginal oasis and reached its peak in the oasis town belt. With the “Kashgar–Aksu–Korla” oasis chain as the main axis, the influence showed gradient differentiation toward the desert transition area. Population growth and contraction corresponded to “development-leading” and “recovery-leading” types of land-use transformation, respectively. (3) Bare land and grassland were the most sensitive links in population–land-use coupling. Bare land showed the pattern of “increasing population and increasing nakedness,” and its sharp decrease was concentrated in towns experiencing population outflow, driven by mining, desertification, vegetation degradation, and other third-party processes. Grassland showed the reverse linkage of “reducing population and increasing grass, increasing population and decreasing grass,” and grassland in population outflow areas was undergoing natural recovery. (4) Population size change and dust-storm PM10 formed two indirect transmission paths in opposite directions through the adjustment of land-use structure. They had a deep impact on regional dust emissions through changes in land cover. Population scale changes shaped the land-use pattern through asymmetric paths such as “increasing population and increasing bare land, decreasing population and increasing grass, and increasing population and occupying farmland.” They indirectly affected regional dust and the atmospheric environment through sensitive land types such as bare land and grassland.

Traditional dwellings in southern Xinjiang, exemplified by the Suohema House, have evolved as adaptive responses to the region’s cold and arid climatic conditions, providing thermally comfortable living environments with relatively low energy consumption. Learning from these climate-responsive design strategies offers an effective approach to reconciling the conflict between energy efficiency and indoor comfort. Such exploration is of great significance for preserving regional architectural identity and promoting the development of low-carbon buildings. This study establishes a performance-driven morphological multi-objective optimization framework for traditional dwellings, taking building energy consumption, thermal comfort, and indoor temperature as the primary optimization objectives. The framework integrates parametric modeling, performance simulation, and multi-objective optimization within the Rhino & Grasshopper platform, employing a genetic algorithm to achieve performance-oriented design exploration. Key design variables were identified through data analysis, and the influence weights and prioritization of morphological parameters were quantified. The results reveal that the room depth in residential dwellings (4.57–4.73 m), room width (3.97–6.75 m), room clear height (2.33–2.42 m), wall thickness (lower wall thickness ranging from 1.14 to 1.22 m, upper wall thickness at 0.76 m), and building orientation (true south) have significant impacts on both energy consumption and indoor thermal performance. Based on these findings, adaptive optimization strategies were proposed from three perspectives: scale optimization, spatial hierarchy refinement, and enhancing the performance of building envelopes. The proposed framework provides methodological guidance for the conservation and adaptive renewal of traditional dwellings, as well as for the design of new, green, and low-carbon residential buildings suited to the climatic conditions of southern Xinjiang.