Applying Time Frequency Analysis and Independent Component Analysis to understand the signals of Groundwater System: A Case Study of Zhuoshuei Plain in Taiwan

The groundwater heads are the response of the aquifer to the external stimuli and are therefore a mixed signal. In order to identify the type of the natural stimuli (i.e. rainfall and river stage), a spatial temporal analysis approach, independent component analysis, was used to analyze the head time series collected from the groundwater-monitoring well network. In order to identify the physical meaning of the independent components (ICs), the derived ICs were compared with the observed river stage and rainfall time series based on the cross correlation, coherence, fast Fourier transform, and wavelet transform methods. The proposed method was applied in the Zhuoshuei River Alluvial Fan to analyze the groundwater levels collected at the 58 observation wells during 2005-2014, and the rainfall and river stage are collected for the comparison analysis as well. The results show that there are two ICs represents the features of rainfall, and two ICs represent the features of river stages. The study results will be the foundation to analyze the pumping features of the study area in the near future.