Evaluating the efficiency of autoregressive models in predicting the vulnerability of coastal aquifers to seawater intrusion

The management of coastal aquifers presents a significant global challenge, primarily due to the threat of seawater intrusion (SWI). This phenomenon, often exacerbated by excessive groundwater extraction, sea-level rise, and climate change, leads to the salinization of freshwater resources, rendering them unsuitable for agricultural, industrial, and domestic use. Consequently, monitoring and predicting the vulnerability of these aquifers is critical for sustainable water resource management and the development of effective mitigation strategies. A key approach for assessing the status of coastal aquifers involves the use of specialized classification indices that quantify their susceptibility to salinization. This study focuses on assessing the vulnerability of a coastal aquifer system to seawater intrusion by employing the Seawater Intrusion Vulnerability Index (SIVI). The SIVI framework was selected for its comprehensive nature, as it distinctively integrates both inherent (static) and dynamic (temporal) characteristics of the aquifer. Inherent parameters, such as hydraulic conductivity and aquifer thickness, define the aquifer's natural buffer capacity, while dynamic parameters, including groundwater level reflect the seasonal and anthropogenic stresses acting upon the system. This dual consideration provides a more holistic and realistic assessment of vulnerability compared to indices that rely on a single category of parameters. The application of the SIVI index to the study area revealed that a substantial portion, approximately 60%, is classified as vulnerable to seawater intrusion. This finding indicates a pressing need for immediate attention and proactive management in these zones to prevent further degradation of water quality. To transition from a static assessment to a predictive outlook, a general sensitivity analysis was conducted to identify the most influential variable governing salinity changes. From a set of potential hydrogeological factors, the concentration of Chloride (Cl⁻) ions was determined to be the most effective and sensitive indicator for tracking the progression of seawater intrusion, given its role as a primary constituent of seawater. Subsequently, a time series forecasting approach was adopted using the AutoRegressive Integrated Moving Average (ARIMA) model. The historical chloride concentration data from 59 observation wells were used to build and calibrate the forecasting model based on sensitivity analysis results. After a rigorous process of model identification, parameter estimation, and diagnostic checking, the ARIMA(2,1,1) model was identified as the most appropriate and robust model for predicting future chloride levels in the region. This specific model configuration implies that the forecast relies on two previous time-lagged values (autoregressive component), one degree of differencing to make the data stationary, and one lagged forecast error (moving average component). The results from the ARIMA(2,1,1) model forecast a concerning trend for the short-term, one-year outlook. The model predicts that the overall vulnerability of the region to seawater intrusion is projected to increase by approximately 8%. The performance of the ARIMA model for this one-year forecasting horizon was evaluated as favorable, demonstrating a Mean Absolute Error (MAE) of 25 mg/L, which is considered acceptable given the context of the study and the variability inherent in hydrogeological data. In conclusion, the integrated methodology of SIVI and ARIMA provides a powerful toolset for both diagnosing current vulnerability and anticipating future risks. The study successfully establishes that the aquifer is not only currently vulnerable but that this vulnerability is on a trajectory to worsen in the immediate future. The insights gained from this research offer a scientifically-grounded basis for implementing targeted monitoring programs, designing controlled extraction policies, and planning strategic interventions to safeguard the long-term sustainability of the coastal aquifer's groundwater resources.