Accurate soil moisture observation is important to the risk assessment, monitoring and early warning of geological disasters. The development and maturity of wireless soil moisture sensor network technology provides the possibility for real-time dynamic monitoring of soil moisture, and can better monitor in-situ data and provide high precision information, which is a good choice to replace the traditional stations. However, the scientific arrangement of those soil moisture sensors is still a big challenge. Based on the study of spatial and temporal variability of soil moisture in Wuhan urban region from year 2010 to 2019, and superimposed with different types of auxiliary data, the study area was divided twice, and an optimal layout method of soil moisture observation network was designed. Firstly, the original Voronoi partition of the region is obtained by using the 32 existing meteorological stations in the region. Based on the data of land use type, the zoning was adjusted, and 16 new zoning were added. Secondly, the results of spatial variability over many years were used to group and classify the regions. The clustering results were used to divide the regions again, and 46 new regions were added. Thirdly, combining with the results of time variability over the years, the optimal location of the station in different partitions (the point with the time root mean square error value closest to 0) was determined, and a new sensor was added for each partition. Finally, based on the original stations, the positions of the new sensors were adjusted, and the redundant locations were deleted. On the basis of the existing 24 stations, 79 new stations were added to the designed observation network, which reduced the existing single point control area to 381~792 km2 in accordance with the requirements of the national code, and the single point representativeness increased by 71.57% on average compared with that before the layout. This paper follows the idea of "partition before laying out", first utilizing the relative continuous satellite remote sensing data to acquire regional soil moisture geography law, and then deduce the layout plan of the ground station network, which can provide a new reference for the optimization of the layout of the related station network, and can be used to monitor the process of regional geological and environmental disasters.