Salinity modelling accuracy of a coastal lagoon: a comparative river flow analysis of basin model vs. traditional approaches

Abstract

The main purpose of this study is to investigate the uncertainties in the modelling of salinity fields in the Ria de Aveiro lagoon associated with the estimation of river flow discharges. The prediction of fresh water inputs is necessary to properly implement forcing conditions and consequently to provide accurate forecasts of baroclinic circulation in coastal lagoons. Located in the north-western Portuguese coast, the Ria de Aveiro is a shallow vertically homogeneous mesotidal coastal lagoon with a complex geometry. Although it is tidally dominated, it receives freshwater from five rivers, the Vouga, Antuã, Cáster, Boco and Ribeira dos Moínhos, whose contributions are responsible for the salinity variation within the system. This research concerns the accurate prediction of river flow to be used in the operational forecast system of the lagoon. Given the lack of observed data for river discharges, as there are only two real time measuring stations located in the Vouga and Antuã river basins, but far from the lagoon, alternative estimation approaches are needed. In order to estimate the river discharges for all five rivers, two different approaches were considered: the first estimates the Vouga river flow, the major fresh water source, based on the nearest real time measuring station and estimates the other river flows based on river basin areas proportionality; the second establishes river flows based on the precipitation/river flow relationships for the Vouga and Antuã rivers and on the areas of the other river basins using the SWAT model. The methodology comprises the exploitation of the 3D unstructured-grid hydrodynamic model SELFE, required to adequately simulate the flow and transport of salt in very complex domains such as the Ria de Aveiro. The model is forced by water elevations at the ocean boundary and river flows at the river boundaries, and atmospheric drivers at the surface (wind stress, atmospheric pressure and heat fluxes). The salinity model predictions were compared with data from seven stations, and its accuracy was assessed through Root Mean Square Error (RMSE). The river flows estimated by the first method led to the best fit between observed and predicted salinity.