Assessment of chemical water types and their spatial variation using multi-stage cluster analysis, Queensland, Australia

A multivariate assessment has been adapted to the classification of a large, irregular dataset of approximately 34,000 surface water samples accumulated over more than 30 years. A two-stage K-means clustering method was designed to analyse chemical data in the form of percentages of major ions (Na, Mg, Ca, Cl, HCO3 and SO4); the first stage of clustering produced 347 groups, which were then re-clustered to generate the final nine water types. The analysis enabled the definition of provinces of water composition and highlighted natural processes influencing the surface water chemistry. On a statewide basis, sodium is the dominant cation and around 50% at all stream flows, while proportions of calcium and magnesium are about equal. Chloride and bicarbonate constitute the bulk of anions present, while sulfate occurs occasionally and tends to be localised. On a global basis, Queensland surface waters are relatively high in sodium, chloride and magnesium, and low in calcium and sulfate. It was also found that the geographical location has a greater impact on major ion ratios than does the stage of stream flow. The regional chemical trends are consistent with geology and climate. Streams in northeast Queensland, with short, steep catchments and high rainfall, yield low salinity, sodium-dominated water; this is also the case for sandy southern coastal catchments. Both also reflect an oceanic influence. The proportions of sodium and chloride decrease westward; streams draining the western side of the Great Dividing Range or flowing into the Gulf of Carpentaria have low salinity but relatively hard water. Streams in western Queensland are higher in calcium and bicarbonate. In the large catchments flowing from Queensland into central Australia, the water composition is highly variable, commonly with elevated sulfate. Also in Queensland, there are several other clearly definable water provinces such as the high magnesium waters of basaltic areas. The findings of this study confirm that the application of such analytical methods can provide a useful assessment of controls over water composition and support management at regional level; the approach used is shown and are applicable to large, disparate datasets.