Regional ground water interpretation using multivariate statistical methods

Gautam Kumar Agrawala, University of Texas at El Paso


The multivariate tools of Principal Components Analysis (PCA) and K-Means Cluster Analysis (CA) were applied analyze the evolution of groundwater chemistry in the El Paso, Texas region. The source data consisted of public domain chemical analyses from 7687 wells. The data was filtered and scanned to remove inconsistent analyses, remove statistical outliers, and determine analytes where data completeness was sufficiently great to allow for multivariate statistical analysis. ^ The data was divided into five basins: (a) Mimbres Basin, (b) Tularosa Basin, (c) Hueco Basin, (d) Rio Grande Basin, (e) Diablo Plateau Aquifer and analyzed in its entirety. For each basin a different subset of analytes was chosen for multivariate analysis depending upon data completeness. The results are presented as: (a) a Digital Elevation Model (DEM) showing all well locations, (b) correlation matrix of all analytes showing all statistically significant linear correlations, (c) table of PCA factor loadings, (d) the DEM illustrating hydrochemical facies derived from PCA and CA, (e) a correlation matrix showing all analytes and the PCA factors, (f) a series of Piper Plots, one for each of the hydrochemical facies, (g) a three dimensional plot showing the hydrochemical facies in relation to the PCA factors, and (h) a series of biplots relating the hydrochemical facies to the loadings for individual analytes. ^ It was hypothesized that modern Geographical Information System (GIS) and multivariate statistical analysis tools can reveal information about the evolution of water chemistry, specifically (a) groundwater flow paths, (b) areas of recharge and discharge, and (c) aspects of the geochemical evolution of groundwater. For this dataset, the multivariate analysis was unable to reveal flow paths, gave partial information on recharge and discharge, and was successful in describing the evolution of groundwater chemistry. The statistical methods were effective at dealing with incomplete data and indicated that the total dissolved solids in the region typically result from dissolution of sodium sulfate rather than evaporative concentration. ^

Subject Area

Hydrology|Engineering, Environmental

Recommended Citation

Agrawala, Gautam Kumar, "Regional ground water interpretation using multivariate statistical methods" (2007). ETD Collection for University of Texas, El Paso. AAI3294433.