Multivariate analysis applied for study of the sampling frequency and the number of sampling stations in water quality monitoring

ABSTRACT

The core of this work consisted of applying multivariate exploratory analysis tools to evaluate the number of water quality monitoring stations and the sampling frequency. In such way, the database provided by the Minas Gerais Institute of Water Management (IGAM) on the river basin of Rio das Velhas, in the most populous central region of Minas Gerais state, was used. The Principal Components Analysis and the Kohonen neural network techniques were applied, resulting in a significant reduction in sampling frequency, in some cases from monthly to annual or semi-annual, and in the reduction of the number of monitoring stations from 36 to 33. The results open the possibility of using these methods as watershed water resources management tools aimed at optimization of water quality monitoring programs.

Keywords:
water quality monitoring; multivariate analysis; principal component analysis; Kohonen neural network