Independent component analysis for characterization and quantification of regional groundwater pumping

Summary The objective of this study is to develop a method that combines signal analysis with a groundwater simulation model for the characterization and quantification of regional groundwater pumping. We first use independent component analysis to identify the characteristics of the main pumping types from hourly groundwater head observations. Then we fit the drawdown variation at each observation well caused by each pumping type with a calibrated groundwater simulation model. We then use the results to estimate the quantity of each pumping type. We apply the proposed method to a regional aquifer system in Sijhou Township, located in the Jhuoshuei River alluvial fan in central-western Taiwan. The independent component analysis identifies three main pumping types: Agricultural type, Industrial type A, and Industrial type B. All three pumping types have the same sine-wave shape with one-day frequency, but differ in peak time, valley time, and peak-to-valley amplitude. Agricultural pumping occurs during daytime farming hours, while industrial pumping occurs during off-peak hours to take advantage of much lower nighttime electricity rates. Quantitative results indicate that the total amount of groundwater pumping in Sijhou is 37,565 tons per day, of which more than 58% is for agricultural use, while the remaining 42% is for industrial use. The percentage of industrial pumping is higher than expected, which may imply that industry use of groundwater is changing. The results indicate a large amount of pumping takes place at depths of more than 40 m. For further investigation, the installation of more monitoring wells is recommended at deeper aquifers.