On the Possibility of Predicting Fresh Submarine Groundwater Discharge Using Global-Scale Datasets of Hydrogeologically Influential Variables Through Regression Modelling and Darcy Velocity Calculations
Abstract
We assess the potential of using global-scale datasets to predict the flux of the fresh component of submarine groundwater discharge (FSGD) along any point along the global coastline. FSGD flux values for unique field sites across the globe were compiled from the literature and were then correlated with data from global-scale datasets of topographic slope (S), saturated hydraulic conductivity (Ksat), and mean annual precipitation (P) for their respective locations. These data were used in two methods to model FSGD flux. The first method was regression modelling where we attempted to model field-derived coastline-normalized FSGD fluxes (m3 m -1 yr-1) with regressions trained by the compiled FSGD flux field data (dependent variable) and the global-scale S, Ksat, and P data (independent variables). We performed univariable regressions of FSGD flux against S, Ksat, and P individually. We then performed multivariable regressions trained by the field-derived FSGD flux data with the global-scale S and Ksat data. In the second method, Darcy velocity calculations of FSGD (cm d-1) using global datasets of S (EarthEnv Global Topography from Amatulli et al., 2018), as a proxy for hydraulic gradient, and Ksat (HiHydroSoil from FutureWater) were performed. We then compared the results of the Darcy calculations with the FSGD flux field data to assess the accuracy of the calculations. The results of the regression modelling were inconclusive as no clear trend could be drawn between FSGD flux and the independent variables. This is likely due to at least three factors: (1) the FSGD flux data sample size is too small (<30 for the multivariable regression); (2) there is too much spatial, temporal, and geological variability between the FSGD field data sources; and (3) there is a need for additional predictor variables (e.g. soil moisture, tidal amplitude, etc.). The results of the second approach using the Darcy calculation were more promising. Our calculated FSGD flux values and the corresponding FSGD flux field data were positively correlated, although the trend did not follow a 1:1 relationship. Further work is needed to increase the sample size of the compiled FSGD flux data. Moreover, for regression, there is a need to identify and incorporate additional predictor variables of relevance that have data available on the global scale.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMOS041..06H
- Keywords:
-
- 1836 Hydrological cycles and budgets;
- HYDROLOGY;
- 1839 Hydrologic scaling;
- HYDROLOGY;
- 1847 Modeling;
- HYDROLOGY;
- 4217 Coastal processes;
- OCEANOGRAPHY: GENERAL