An objective water quality index.

Pollution Control Act Amendments of 1972, several inquiries about how to com bine the numerous water quality param eters have been received. The desired summarization should be concise and ob jective and should inform administrators clearly of water quality trends. In gen eral, graphs of parameters against each other or against time are not concise and do not show trends clearly because of data overlap and sheer volume. The index developed by Brown et al.1 is not really objective because a panel of "experts" rates the water quality param eters to be used. Many users feel that dif ferent panels will give different ratings, thus destroying comparability and objec tivity. A technique developed by Kendall,2 which is relatively unknown to data ana lysts in the water field, has the desired properties of objectivity, conciseness, and ease of application. Basically this method, which will be demonstrated below, is a distribution-free statistical procedure for mapping p-measurements on an object onto a line so that existing differences in location in p-space are preserved and gen erally more clearly defined. Seasonal trends existing among a group of param eters is generally enforced by use of the following method. The only hindrance in using this method is the fact that the num bers generated in one evaluation cannot be directly compared with those generated by a different run. The two sets of raw data must be combined and a new evalua tion of the total data set made. Method There are many data evaluation methods available for detecting trends. These in elude graphs, multivariate analyses, re gression analyses, and "expert" judgment. Frequently the number of charts and graphs required to evaluate data is equal to or greater than the volume of the raw data, thus largely defeating the intent of clarifying a situation. Multivariate and regression analyses are not appropriate because the basic assump tions of multivariate normality, homogene