Time-frequency analysis of groundwater depth variation based on the ICA-WTC composite method

Abstract

Investigating the impact and response of precipitation on groundwater resources is critical to groundwater resource management because groundwater is primarily recharged by precipitation. The large-scale climatic patterns and climate change affecting groundwater resources and their response through precipitation are poorly understood. Therefore, we propose a systematic method of time–frequency correlation analysis combining independent component analysis (ICA) and wavelet coherence analysis (WTC). This method derives independent components (ICs) with precipitation features from groundwater depth to obtain more information from the time–frequency perspective and prove its reliability and superiority compared with traditional methods. ICs with precipitation features (ICprecipitation) with teleconnection factors of large-scale climatic patterns were analyzed in time–frequency, revealing the effects of different patterns on groundwater depth changes through precipitation. This result demonstrates the potential of the ICA-WTC method for research on climate-precipitation-groundwater physical response mechanisms. Compared with the traditional method, this method removes the non-precipitation influence in the groundwater depth variation to obtain the independent component factors with precipitation influence features in the groundwater depth variation, this process demixes the groundwater depth response to all external influences to achieve an accurate extraction of the groundwater response to the precipitation influence, so that the analysis can be specific to the influence and response process of groundwater and precipitation; for the analysis results, it uses a range to describe time lag of precipitation influenced groundwater instead a fixed value, and the method visualize the time–frequency response process of the large-scale climatic patterns affecting groundwater through precipitation and provides a basis for further analysis. This method was applied to the typical area of Sanjiang Plain in Heilongjiang Province, Northeast China, one of the largest agricultural areas in China and one of the areas with the fastest groundwater decline due to groundwater mining for agricultural irrigation. Three ICprecipitation of groundwater depth were obtained, representing the annual period, long-term period, and both the annual period and long-term period of precipitation influence, with time lags of precipitation influence on groundwater depth from 63.02 ± 28.00 day to 176.99 ± 28.63 day. The WTC of the four large-scale climatic patterns located in different directions from the study area with ICprecipitation shows that the large-scale climatic patterns affect groundwater mainly by influencing the long period of precipitation. Moreover, the El Niño Southern Oscillation pattern still is the most influential. Even though the study area is located in the temperate zone of the Northern Hemisphere, it is still more vital than other models closer to the study area.