Enhancing River Discharge Monitoring through Information Theory: Insights from the Surma River

Abstract

In developing countries such as Bangladesh, river discharge monitoring networks are frequently designed inefficiently, resulting in limited spatial and temporal coverage. They are operated poorly, and they often fail to achieve their purposes, resulting in the unavailability of data to describe the behavior of such systems. In these cases, water-related decisions may create problems for the environment, the regional economy, and society. This paper investigates the application of Shannon’s information theory to the design and evaluation of an efficient discharge monitoring network for the Surma River. A one-dimensional (1D) model was formulated in order to extract all discharge data for different points of the Surma River by using MIKE 11. The appropriate monitoring station locations were determined by optimizing two conflicting objective functions (by maximizing the joint entropy and minimizing the total correlation) using the Nondominated Sorting Genetic Algorithm II (NSGA-II) and a greedy algorithm. The study demonstrates that an informative and less redundant monitoring network configuration can be identified through the NSGA-II. For the Surma River, the use of a discharge data set that includes extreme flow values does not affect the optimal number of monitors and results in similar optimal placement options.