Orthogonal Variance Structures in Lake Water Quality Data and Their Use for Geo-chemical Classification of Dimictic, Glacial/Boreal Lakes

The accumulating volumes of data collected within environmental monitoring programs facilitate the use of exploratory statistical methods of data analysis as a supplement to traditional methods of characterizing lake water quality. When principal component analysis and multidimensional scaling are applied to a matrix containing approximately 24000 samples of lake water quality variables pH, alkalinity, conductivity, hardness, color, Secchi depth and total phosphorus concentration, it is found that the total matrix variance can be approximately reproduced in an orthogonal two-dimensional base with transformations of hardness and color as best principal component representatives. This base is suggested as an empirical lake classification standard where the variance structure of subset lake populations (such as single lakes) can be referenced to the water quality standard of the generic population. Since the principal axes of the base exclusively contain inorganic and organic related variables respectively, the combined inorganic/organic characteristics of the lake can be expressed with the hardness and color variables alone. With the data matrix being large enough to produce high significance levels, and with variable ranges wide enough to represent a majority of dimictic, glacial/boreal lakes, the analysis results should be valid in many lakes throughout the world.[PUBLICATION ABSTRACT]