River temperature analysis with a new way of using Independant Component Analysis

Abstract

In the context of global warming, river management is essential to maintain favourable water temperature ranges for aquatic species. Therefore, understanding the main factors influencing the water temperature becomes a key part in the management process. In this paper, we used Independent Component Analysis (ICA) to identify these main factors and improve water temperature forecasting. The study is caried out on two rivers in Normandy (France) with quite different characteristics. Each river was equipped with several temperature sensors which series range from 2011 to 2021. The ICA analysis of the data series reveals that the thermal regime of these two rivers is mainly controlled by seasonal and daily climatic factors. The Sélune regime also turns out to be influenced by the presence of a dam, dismantled during the monitoring of the river. The temperature of the Odon appears to be clearly controlled by seasonal lightening conditions in connection with the presence of the riparian vegetation. Complementary, an innovative approach called “successive ICA” is used to reconstruct the natural thermal regime of the Sélune without the presence of the dam. Emphasis is therefore placed here on the interest of ICA in hydrology as en elementary method for extracting the main influencing factors and quantifying their importance on the thermal regime of a river. It also allows to remove the influence of a particular factor and reconstruct time series better suited for temperature forecasting. The method used here is not specific to temperature time series and can be applied to any region even with different hydrological characteristics.