Factor analytical study on water quality in Lake Saimaa, Finland

Abstract

Multivariate data analysis methods (4-way Candecomp-PARAFAC model solved with Multilinear Engine (ME-1)) were used to interpret the data of over two decades to study the changes in the water of Lake Saimaa in Finland. Earlier studies have shown that it is difficult to extract the natural background from the other sources of variation. By using the multilinear model three interpretable factors representing natural and anthropogenic processes could be extracted. The natural long-term variation, seasonal fluctuation and dilution of discharges in the recipient area could be extracted into their own factors, which could be easily visualized. The variation could be also presented with estimated variation in the water quality parameters caused by each of these natural or anthropogenic processes.