Two-decade variations of fresh submarine groundwater discharge to Tolo Harbour and their ecological significance by coupled remote sensing and radon-222 model

Abstract

Although submarine groundwater discharge (SGD) comprises an insignificant proportion of the global hydrologic cycle, it contributes significantly to chemical fluxes into the coastal waters due to concentrated constituents in coastal groundwater. Large nutrient loadings derived from SGD can lead to a series of environmental and ecological problems such as algal blooms, resulting in water discoloration, severe dissolved oxygen depletion, and eventually beach closures and massive fish kills. Previous studies have demonstrated the relationship between algal blooms and SGD obtained from direct measurement with seepage meters or from geo-tracer (i.e., radon and radium) based models; these traditional methods are time-consuming, laborious and point monitoring, and can hardly achieve a high spatiotemporal resolution SGD estimation, which is vital in revealing the effects of SGD to algal blooms over a long period. Alternatively, remote sensing methods for high spatiotemporal resolution SGD localization and quantification are applicable and effective. The temperature difference or anomaly between groundwater and coastal water extracted from satellite thermal images can be used as the indicator to localize and detect SGD especially its fresh component (or fresh SGD). In this study, multi-year (2005, 2011 and 2018) radon samples in Tolo Harbour were used to train regression models between in-situ radon (Rn) activity and the temperature anomaly by Landsat satellite thermal images. The models were used to estimate two-decade variations of fresh SGD in Tolo Harbour. The synergistic analysis between the time series of fresh SGD derived from regression models and high spatiotemporal resolution ecological metrics (chlorophyll-a, algal cell counts, and E.coli) leads to the findings that the increase of the fresh SGD associated with high nutrient concentrations is witnessed 10–20 days before the observations of algal bloom events. This study makes the first attempt to demonstrate the strong relation between the SGD and algal blooms over a vicennial span, and also provides a cost effective and robust technique to estimate SGD on a bay scale.