GENERALIZED WATER-QUALITY INDEX BASED ON MULTIVARIATE FACTOR-ANALYSIS

Water quality indexes permit an evaluation of water quality conditions in comparative terms. This paper presents a methodology for derivation of a generalized water quality index based on factor analysis (a multivariate statistical technique), and examines the geographical applicability of such an index. Ten pollutant parameters from water quality data accumulated in the Carson Valley of Nevada during the 1974, 1975, and 1976 irrigation seasons provided the basis for 2 indexes: (1) WQITN = f(temp., BOD, TP, EC, DODP); and (2) WQIPN = f(temp., BOD, PO4-P, EC, DODP). Regression analysis of the indexes as functions of water quality variables from 2 river basins in Nevada indicated WQITN to be the better index (F-test 99%, R2 = 0.9098). The WQITN formulation was then compared with index values suggested by water quality experts (WQIE). Regression coefficients were determined for WQIE and WQITN index regression equations as functions of water quality data from selected locations throughout the United States. The proposed WQITN appeared the most geographically acceptable index (F-test 99%, R2 = 0.9754). The model estimated by factor analysis was shown to have good potential as a generalized water quality index. Use of the suggested methodology for the development of generalized, as well as specific-use water quality indexes, may enable better geographic identification of problem areas contributing to water pollution and the development of appropriate water quality standards. Standardization of the indexes in terms of a range of acceptability, e.g., light, moderate, or severe pollution, is suggested for further investigation.; This paper presents a methodology for derivation of a generalized water quality index based on factor analysis (a multivariate statistical technique), and examines the geographical applicability of such an index. Ten pollutant parameters from water quality data accumulated in the Carson Valley of Nevada during the 1974, 1975, and 1976 irrigation seasons provided the basis of two indexes: WQITN equals f(temp. , BOD, TP, EC, DODP); and WQIPN equals f(temp. , BOD, PO//4-P, EC, DODP). Regression analysis of the indexes as functions of water quality variables from two river basins in Nevada indicated WQITN to be the better index. The WQITN formulation was then compared with index values suggested by water quality experts (WQIE). Regression coefficients were determined for WQIE and WQITN index regression equations as functions of water quality data from selected locations throughout the United States. The proposed WQITN appeared the most geographically acceptable index. The model estimated by factor analysis was shown to have good potential as a generalized water quality index.