Optimal rain gauge network design based on multi-objective optimization approach

Abstract

Precipitation is a mam input to many hydrological applications, such as water management, flood forecastmg and hydrological modelling. The goodness of the ramfall field estimation can thus affect their performances. Despite radar-based and satellite-based measurements have nowadays become very common and accurate, ram gauges monitoring stations are still needed. The gauge density and its spatial distribution are two of the key factors influencing the accuracy in precipitation estimation. Even if in the last decades many studies proposed several methodologies for the design of optimal monitoring networks, only few studies use hydrological model performance as a design criterion. The purpose of this study is to define the optimal ram gauge network for the Mignone River catchment (Italy). The optimal network is defined through a multi-objective optimization approach, where the interpolation error of precipitation is minimised and the performance of a hydrological model based on the Width Function Instantaneous Unit Hydrograph theory is maximised. The optimization is run both without and with constraints, which are based on rainfall patterns. A score to choose the best set of points in the Pareto front is presented. The results suggest that there are preferential areas where sensors locations achieve optimal interpolation error and model performance.