Statistical modelling of water quality time series: the River Vouga basin case study

Abstract

The focus of this work is on contributions to the analysis of time series of water quality variables. The study of these time series has characteristics related to data collection procedures, as well as to several economic and geographic aspects. Some characteristics of water quality variables pose challenges to their statistical modelling, namely, seasonality, change-point detection, the existence of outliers, etc. Some of these topics will be addressed in this chapter. The presentation of these topics is accompanied by the study of water quality data from the hydrological basin of the Vouga River and the Ria de Aveiro lagoon. In particular, statistical modelling is performed using data pertaining to the dissolved oxygen concentration variable (in mg/l) from the Carvoeiro water-monitoring site in the Vouga River basin. Section 2 discusses the area under study, while Section 3 presents a data descrip‐ tion with an exploratory analysis of the primary monitoring sites of the Vouga basin. Section 4 addresses the analysis of a time series through the decomposition into its main components and their treatment, while Section 5 presents the modelling of time series through a state space model approach.