Studying the temporal and spatial variability of soil moisture in artificial ecosystems is of great significance for guiding the restoration of vegetation patterns and the efficient management and utilization of regional water resources. We selected an artificial vegetation restoration slope with a depth of 0-200cm in the loess alpine region for research, in order to reveal the spatio-temporal variation characteristics of its soil moisture. During the vegetation growth peak season in 2018 (June-August), continuous dynamic data of soil moisture content in different soil layers were measured by neutron meter. Based on this, the spatio-temporal variability and temporal stability of soil moisture at different depths were studied by using classical statistics and temporal stability analysis. The results showed that there was no significant difference in the soil moisture content of each soil layer profile during the period of measurement. At the spatial scale, the result indicated that all the soil layers showd a moder variance which increased with the increase in soil depth. At the time scale, the topsoil layer showed a moderate variance while the other layers exhibited weak variance, which suggested that time variability of soil moisture in deep layer was smaller than that in shallow layer. The spearman rank correlation coefficient of soil moisture content of 0-200cm in the experimental site reached 0.8 or more at different measuring periods, and the correlation with moisture was extremely significant. As the depth of the soil increased, the temporal stability of soil moisture also increased between different soil layers. Based on relative difference analysis, the representative measuring points can be applied to predict the average soil moisture content of the corresponding soil layers in a region (determination coefficient R² is 0.7138-0.8605). In conclusion, with this paper we aim to provide a theoretical basis for the deployment of soil moisture monitoring sites, and suitable guidance for the selection of vegetation restoration and ecological reconstruction models.