Global Regression Modelling of Fresh Submarine Groundwater Discharge (FSGD) Using Multivariable Analysis of Aquifer Characteristics

It has been demonstrated that submarine groundwater discharge (SGD) is a common phenomenon along the global coastline and can impact nearshore biogeochemistry. The most common techniques for quantifying SGD are field-based methods such as seepage meters and radioisotope tracers. However, these methods are limited in geographic scale, are laborious, and can be costly. Numerical modelling offers a liberating alternative to field-based methods of SGD quantification. This research presents the foundation for a novel modelling technique for the characterization of the fresh component of SGD using multiple linear regression analysis. Training data for this new model, the Nutrient Export Through Fresh Submarine Groundwater Discharge Model (NExT FSGD), is organized in two categories. The first category is coastline-normalized FSGD flux data for study sites around the globe that are reported in published literature. The second category is hydrogeological parameter data from global-scale datasets that are associated with the FSGD flux data using the flux data’s geographic coordinates. FSGD flux data for approximately 40 unique sites have been compiled. Study sites are categorized into six types of aquifer systems: unconsolidated, sedimentary rock, volcanic rock, carbonate/karst, crystalline (metamorphic and plutonic rocks), and mixed. Data from three global-scale hydrogeological parameter datasets that provide porosity, permeability (log(k)), precipitation (mm/yr), and topography (degrees) data are being extracted. The goal of this research is to create a multivariable statistical equation using these four hydrogeological parameters (independent variables) as predictors of FSGD flux (dependent variable) at any point along the global coastline. This tool could be used to help predict FSGD-sourced nutrient fluxes to the ocean. Preliminary results that sought to find a correlation between FSGD flux and permeability were unsuccessful. One possibility for this may be a need for more FSGD flux data as the sample size was only 35 data points. There could also be a need to further curate the dataset. The next steps for this research include further curating the FSGD flux data, compiling more FSGD flux data to represent a greater breadth of global aquifer systems, and incorporating more hydrogeological parameters.