The cleaning of the atmospheric electric field is the key step of pretreatment, which is of great significance to the subsequent excavation research. In view of the shortcomings of traditional anomaly detection algorithm, which needs to specify the corresponding parameters and fail to use the relevant information between time series, a new outlier detection and correction method based on the combination of isolation forest and Chen-Liu algorithm is proposed. The method uses ARIMA model to combine the atmospheric electric field to get the fitting residual. The isolation forest model is constructed based on residual sequence to determine the location of the outliers. Finally, the Chen-Liu algorithm is used to correct the outliers. The reliability of the proposed method is verified by simulation series and the atmospheric electric field test. Compared with the original prediction, the results of the prediction of the series of thr atmospheric electric field after cleaning are improved by 27.8% and 34.98% respectively in root mean square error and mean percentage error.