EVALUATING AQUIFER-SYSTEM PROPERTIES FROM EXTENSOMETER AND WATER-LEVEL DATA USING PRINCIPLE COMPONENT ANALYSIS

Detailed aquifer characterization is necessary for quantifying sustainable water resources but is often extremely difficult and expensive to perform. However current studies utilizing high precision subsidence data along with water-level data can be utilized as observations to characterize aquifer systems resulting in reliable and highly diagnostic parameter information. In this investigation we apply the statistical method Principle Component Analysis to the subsidence record at a single site along with nested water-level data from the same location to characterize the aquifer and aquitard properties.

More than a decade of hourly subsidence and water-level data from three nested piezometers, open to aquifers at three different depths, have been analyzed from the Lorenzi extensometer site in Las Vegas Valley located near the edge of the northwest subsidence bowl. The data clearly reveal three separate signals; a long-term (decadal) deformation signal, a seasonal signal representing the seasonal pumping within the valley, and a daily signal representing the daily municipal pumping cycles. Within the seasonal signal a hydrodynamic lag can be detected as well.

Principle Component Analysis is used to identify the significant factors causing each of the signals as well as the degree of each factor’s influence on the behavior of the signals. The method is also utilized to determine the relationship between the fluctuating head levels and subsidence within the seasonal signal. By gaining a better understanding of the existing data, calculations are performed to obtain values for not only the aquifer properties but also the elastic and inelastic components of the aquitard properties. Principle Component Analysis precludes the need for more complex numerical inverse models for determining these important aquifer system properties, which cannot typically be attained from single extensometer records.