Based on data from multiple sources including Landsat8 OLI_TIRS remote sensing images, digital elevation model (DEM), road networks, and water system, as well as the corrected energy factor model and gravity model, this study improved the method for ecological network construction based on morphological spatial pattern analysis (MSPA) and the minimum cumulative resistance (MCR) model. Using the method, this study constructed and optimized a hierarchical ecological network for Zhongwei City, Ningxia. The results show that: ① The identification of ecological source areas in arid regions should be conducted using the data of woodland, grassland, and water bodies as foreground data and woodland and selecting ecological source areas based on ecological redlines; ② The ecological network of Zhongwei City presents a spatial structure mode characterized by four cores, three first-level ecological corridors, and multiple minor ecological source areas. Seventeen ecological source areas were identified, accounting for 22.33% of the study area. Among them, four grade-1 and -2 source areas exhibited significantly high energy factors, forming four cores. A total of 33 potential ecological corridors were identified, including three grade-1 ecological corridors; ③ Strategies for optimizing the ecological network, including improving ecological source areas’ quality, enhancing corridors, and restoring ecological breaking points. Except for the stable corridors connecting Nos. 1, 7, and 9 source areas, other source areas manifest poor connectivity, leading to low ecological network stability. Therefore, it is necessary to establish 24 ecological stepping stones. Furthermore, 38 ecological breaking points requiring urgent restoration were discovered; (4) The optimized ecological network demonstrates enhanced stability, with α, β, and γ indices elevated at 9.5%, 3.8%, and 4.2%, respectively. This network will promote the flow of ecological materials/information and biodiversity conservation.