Title: Multivariate Statistical Assessment of Groundwater and Surface Water Quality in Otuabagi Community, Bayelsa State, Nigeria: Implications for Sustainable Water Resources Management
Authors: IJAOLA opololaoluwa Oladimarun
ITUGHA Okuroghoboye Diepreye
Volume: 10
Issue: 8
Pages: 52-62
Publication Date: 2026/08/28
Abstract:
Multivariate statistical techniques provide a means of moving beyond simple comparison of water quality parameters with regulatory standards toward identifying the underlying processes and pollution sources controlling water chemistry. This study applies descriptive statistics, Pearson correlation analysis, Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), and a Multiple Linear Regression (MLR) predictive engineering model to fourteen physicochemical parameters measured in ten groundwater and surface water samples from Otuabagi Community, Ogbia Local Government Area, Bayelsa State, Nigeria. Correlation analysis identified several strong associations (|r| >= 0.70), most notably between electrical conductivity and acidity (r = -0.91), electrical conductivity and dissolved oxygen (r = -0.88), dissolved oxygen and acidity (r = 0.88), and total hardness and chloride (r = -0.87), suggesting shared hydrogeochemical controls linked to mineral dissolution and buffering capacity. Principal Component Analysis extracted four components explaining 87.67% of total variance, with the first two components (64.28% cumulative) dominated by dissolved oxygen, carbonate/bicarbonate, total hardness, and acidity on PC1, and biochemical oxygen demand, magnesium, iron, and total dissolved solids on PC2, interpreted respectively as a geogenic mineral-buffering factor and an organic/anthropogenic loading factor. Hierarchical Cluster Analysis using Ward's method perfectly partitioned the ten samples into two clusters corresponding exactly to source type, groundwater versus surface water, indicating that source type is the dominant control on overall hydrochemical similarity in the study area. An MLR model developed to estimate WQI directly from four rapidly field-measurable parameters (electrical conductivity, pH, total hardness, and total alkalinity) achieved a coefficient of determination (R2) of 0.9648 (adjusted R2 = 0.9366), demonstrating strong potential as a low-cost surrogate estimation tool for preliminary water quality screening ahead of full laboratory analysis. These findings demonstrate that multivariate statistical analysis, combined with a simple predictive engineering model, provides mechanistic insight into water quality variability that complements conventional index-based assessment and offers a scientific basis for targeted, source-specific water resources management in oil-impacted communities of the Niger Delta.