Evaluation of the fast orthogonal search method for forecasting chloride levels in the Deltona groundwater supply (Florida, USA) | Evaluation de la méthode de recherche orthogonale rapide pour la prévision des concentrations en chlorure dans l’approvisionnement en eaux souterraines de Deltona (Floride, Etats-Unis d’Amérique) Evaluación del método de búsqueda ortogonal rápida para pronosticar los niveles de cloruro en el suministro de agua subterránea en Deltona (Florida, EE UU) 美国佛罗里达州Deltona地下水供应中氯离子含量的快速正交搜索方法评估 Avaliação do método da busca ortogonal rápida para a previsão de níveis de cloreto no fornecimento de águas subterrâneas de Deltona (Flórida, EUA)

Despite the broad impact and importance of saltwater intrusion in coastal aquifers, little research has been directed towards forecasting saltwater intrusion in areas where the source of saltwater is uncertain. Saline contamination in inland groundwater supplies is a concern for numerous communities in the southern US including the city of Deltona, Florida. Furthermore, conventional numerical tools for forecasting saltwater contamination are heavily dependent on reliable characterization of the physical characteristics of underlying aquifers, information that is often absent or challenging to obtain. To overcome these limitations, a reliable alternative data-driven model for forecasting salinity in a groundwater supply was developed for Deltona using the fast orthogonal search (FOS) method. FOS was applied on monthly water-demand data and corresponding chloride concentrations at water supply wells. Groundwater salinity measurements from Deltona water supply wells were applied to evaluate the forecasting capability and accuracy of the FOS model. Accurate and reliable groundwater salinity forecasting is necessary to support effective and sustainable coastal-water resource planning and management. The available (27) water supply wells for Deltona were randomly split into three test groups for the purposes of FOS model development and performance assessment. Based on four performance indices (RMSE, RSR, NSEC, and R), the FOS model proved to be a reliable and robust forecaster of groundwater salinity. FOS is relatively inexpensive to apply, is not based on rigorous physical characterization of the water supply aquifer, and yields reliable estimates of groundwater salinity in active water supply wells.