Development of a software package for trend detection in temporal series: Application to water and industrial effluent quality data for the St. Lawrence River.

Despite the considerable amount of effort and resources involved in monitoring water quality, water quality assessment and environmental follow-up are sometimes carried out with simple statistics, the main reason being the lack of appropriate statistical methods adapted to the nature of sampled water quality data.A survey of the classical methods used for trend detection and of their limitations is first presented, including the most recent non-parametric techniques adapted to the structure of the sampled data and to the possible types of trends occuring. This paper then presents an interactive user-friendly software package developed for microcomputers making use of these latest adapted techniques. Afterwards, some applications of the software are described pertaining to the concentrations measured at long-term stations on the St. Lawrence River and to the mass loadings discharged by regulated industries. Finally, conclusions are drawn about the assumptions, performance and limitations of the package as well as about the research needs to improve the usefulness and applicability of the software.