Abstract:Our country has established an improved seismic underground fluid observation network.Underground water level is an important earthquake precursor factor and is applied to prediction measures.The observation data of groundwater level is an integrated parameter and includes many components determined by such factors as precipitation,atmospheric pressure,Earth tide,and geological structures.To highlight the influence of tectonic forces on groundwater level and to recognize the earthquake precursorinformation obtained by this parameters must be separated from original observation data.Components influenced by Earth tide and atmospheric pressure can be separated on the basis of a widely used theory.The influence of climate conditions on underground water level at a watershed scale,changes of which are long-term and periodic,can be separated through smooth processing methods.On the contrary,rainfall peaks influence underground water level,which reduces the effects created by structure forces;therefore,the identification of earthquake precursor information is difficult.In order to highlight tectonic forces on groundwater level,it is necessary to separate the peak value of rainfall effects on underground water level.In our study,we regarded the observation data of groundwater level as a linear combination of all of the components.Changes caused by Earth tide and atmospheric pressure were first separated according to the BAYTAP-G procedure.Then,because quick and slow response existed in the influence of rainfall on groundwater level,these effects were distinguished by applying base flow separation methods.Changes induced by rainfall were separated from groundwater level by using base flow separation methods.By analyzing the relationships between the curve feature of every component and local seismic activities,local earthquake precursor information on groundwater