Abstract. Due to the complex geological structure of landslides, the installation of a monitoring network could be useful for a variety of scopes studying the possible evolution of a landslide for early warning, and the occurrence of disasters of different types landslides is different not only in the form of deformation, but also in the trigger factor. In the process of landslide monitoring, due to equipment failure and external factors, data loss or abnormal are inevitable. In this paper, through the processing and analysis of the monitoring data of the Zhutoushan landslide, the landslide is rotational landslide which is caused by the rainfall. The box plot is used to detect outliers, and the polynomial fitting function and the moving average denoise method are compared to repair the data, and the latter is better. Through the exploratory analysis of GNSS data, the correlation between monitoring points at different locations is found, which provides a basis for the identification of landslide types.