Statistical study to identify the key factors governing ground water recharge in the watersheds of the arid Central Asia.

Understanding the source and recharge of ground waters is of great significance to our knowledge in hydrological cycles in arid environments over the world. Northern Xinjiang in northwestern China is a significant repository of information relating to the hydrological evolution and climatic changes in central Asia. In this study, two multivariate statistical techniques, hierarchical cluster analysis (HCA) and principal component analysis (PCA), were used to assess the ground water recharge and its governing factors, with the principal idea of exploring the above techniques to utilize all available hydrogeochemical variables in the quality assessment, which are not considered in the conventional techniques like Stiff and Piper diagrams. Q-mode HCA and R-mode PCA were combined to partition the water samples into seven major water clusters (C1-C7) and three principal components (PC1-PC3, PC1 salinity, PC2 hydroclimate, PC3 contaminant). The water samples C1 + C4 were classified as recharge area waters (Ca-HCO3 water), C2 + C3 as transitional zone waters (Ca-Mg-HCO3-SO4 water), and C5 + C6 + C7 as discharge area waters (Na-SO4 water). Based on the Q-mode PCA scores, three groups of geochemical processes influencing recharge regimes were identified: geogenic (i.e., caused by natural geochemical processes), geomorphoclimatic (caused by topography and climate), and anthropogenic (caused by ground water contamination). It is proposed that differences in recharge mechanism and ground water evolution, and possible bedrock composition difference, are responsible for the chemical genesis of these waters. These will continue to influence the geochemistry of the northern Xinjiang drainage system for a long time due to its steady tectonics and arid climate. This study proved that the chemistry differentiation of ground water can effectively support the identification of ground water recharge and evolution patterns.