Multivariate Statistical Analysis of Geochemical Data of Groundwater in El-Bahariya Oasis, Western Desert, Egypt

摘要

The aim of the present study is to study the application of multivariate statistical analyses of hydrochemical data using the chemical analyses for 125 groundwater samples with 18 parameters include the hyrdrochemical compositions (Ca~(2+), Mg~(2+), Na~+, K~+, (HCO_3), (SO_4)~(2-) and Cl~-) and the physicochemical parameters (EC, TDS, TH, SAR, RSBC, PI, KR, SSP, MAR, RSC and Na%). The linear regression is an approach to modeling the relationship between two variables using a set of individual data point and used to explain or predict the behavior of a dependent variable. Two variables were used to develop a relationship between TDS as an independent variable and different hyrdrochemical data as a dependent variable. Using these equations, by known TDS value, the equation tries to predict any unknown other variables. The linear regression equations used also between the EC as an independent variable and all different water quality variables. The correlation matrix performed for the groundwater using the hyrdrochemical compositions (r varies from 0.84 to 0.08). All data have positive relations reflecting direct relationship with all hydrochemical data. Good correlation observed between TDS and each of other variables, while weak positive relation detected between (HCO_3) and Ca~(2+), (SO_4)~(2-), Ng~(2+) and Cl~-. Two clusters were performed, the first use TDS, Ca, Mg, Na, K, HCO_3, SO_4, Cl, EC and TH while the second use PI, TH, MAR, EC, SAR, KR, Na%, RSBC, RSC and SSP as variables. Skewness and kurtosis are calculated for all data to describe the shape and symmetry of the distribution of geochemical data along the study area. Skewness values vary from 3.22 to -1.36. Positive skewness were notice in most parameters indicates that the shape of their statistical distribution diagrams show the tail on the right side (direction of high values) is longer than the left side and the bulk of the values (possibly including the median) lie to the left of the mean for each parameter. Kurtosis values vary from 18.17 (for SO_4) to -0.65 (for RSBC). Positive Kurtosis characterize most parameters indicates a peaked distribution relative to a normal distribution of the data, while the other are negative (indicates a flat distribution). The SO_4, KR, MAR and TH have high kurtosis values, indicates tend to have a distinct peak near the mean and have heavy tails.