Multivariate analysis in hydrology: the factor correspondence analysis method applied to annual rainfall data

Abstract

Most research on hydrology in developed countries does not emphasize problems that are often encountered by hydrologists in developing countries, problems such as the scarcity and poor quality of data. In developing countries the available data base is often a major constraint and a limiting factor in any hydrological study. This paper analyses the consistency of a network of nonrecording raingauges based on the annual rainfall recorded by six representative stations in the Tacuarembô River basin in Uruguay. The identification of possible errors not revealed in the printed records was accomplished by application of the factor correspondence analysis method and verified with success by the classical double-mass analysis. This ability to identify suspicious data with very modest requirements for data management implies that the methodology based on factor correspondence analysis could find wide application.