Abstract: The annual changes of three kinds of deformation data in Dahuichang are obviously related to the changes of rainfall and temperature, the latter disturbs the significant signals in the varied curves to some extent. However, the interferences are far from being enough to obscure the remarkable changes started since 1975 before the Tangshan Earthquake. It is difficult to determine the amplitude and onset of the anomalies because of the existence of the interferring factors due to rainfall and temperature changes.We adopt mulit-channel Wiener filtering method to predict the output of the data (such as shortline levelling, watertube tilt and extensometer) by the input (rainfall and temperature). The difference, between the predicted output and the observed data (prediction error) is the significant signals that we desire.The anomalous change of shortline levelling started at Dahuichang in April, 1975, just one year and a little more before the Tangshan Earthquake. Obviously short-term anomalies had been observed about three months before the earthquake. The total amplitude of anomaly is about 2mm (the base line of the short levelling is 26 meters). The anomalies show that the Pabaoshan fault was thrusting before the Tangshan. Earthquake.