Ho Kyung Ha

Machine learning approaches to coastal water quality monitoring using GOCI satellite data

Since coastal waters are one of the most vulnerable marine systems to environmental pollution, it is very important to operationally monitor coastal water quality. This study attempts to estimate two major water quality indicators, chlorophyll-a (chl-a) and suspended particulate matter (SPM) concentrations, in coastal environments on the west coast of South Korea using Geostationary Ocean Color Imager (GOCI) satellite data. Three machine learning approaches including random forest, Cubist, and support vector regression (SVR) were evaluated for coastal water quality estimation. In situ measurements (63 samples) collected during four days in 2011 and 2012 were used as reference data. Due to the limited number of samples, leave-one-out cross validation (CV) was used to assess the performance of the water quality estimation models. Results show that SVR outperformed the other two machine learning approaches, yielding calibration R2 of 0.91 and CV root-mean-squared-error (RMSE) of 1.74 mg/m3 (40.7%) for chl-a, and calibration R2 of 0.98 and CV RMSE of 11.42 g/m3 (63.1%) for SPM when using GOCI-derived radiance data. Relative importance of the predictor variables was examined. When GOCI-derived radiance data were used, the ratio of band 2 to band 4 and bands 6 and 5 were the most influential input variables in predicting chl-a and SPM concentrations, respectively. Hourly available GOCI images were useful to discuss spatiotemporal distributions of the water quality parameters with tidal phases in the west coast of Korea.

Variability of Warm Deep Water Inflow in a Submarine Trough on the Amundsen Sea Shelf

The ice shelves in the Amundsen Sea are thinning rapidly, and the main reason for their decline appears nto be warm ocean currents circulating below the ice shelves and melting these from below. Ocean currents ntransportwarm densewater ontothe shelf,channeledby bathymetric troughs leadingto the deep inner basins. nA hydrographic mooring equipped with an upward-looking ADCP has been placed in one of these troughs on nthe central Amundsen shelf. The two years (2010/11) of mooring data are here used to characterize the inflow nof warm deep water to the deep shelf basins. During both years, the warm layer thickness and temperature npeaked in austral fall. The along-trough velocity is dominated by strong fluctuations that do not vary in the nvertical. These fluctuations are correlated with the local wind, with eastward wind over the shelf and shelf nbreak giving flow toward the ice shelves. In addition, there is a persistent flow of dense lower Circumpolar nDeep Water (CDW) toward the ice shelves in the bottom layer. This bottom-intensified flow appears to be ndriven by buoyancy forces rather than the shelfbreak wind. The years of 2010 and 2011 were characterized by na comparatively stationary Amundsen Sea low, and hence there were no strong eastward winds during winter nthat could drive an upwelling of warm water along the shelf break. Regardless of this, there was a persistent nflow of lower CDW in the bottom layer during the two years. The average heat transport toward the ice nshelves in the trough was estimated from the mooring data to be 0.95 TW.

Some Implications of Ekman Layer Dynamics for Cross-Shelf Exchange in the Amundsen Sea

Theexchangeofwarm,saltyseawateracrossthecontinentalshelvesoffWestAntarcticaleadstosubsurface glacial melting at the interface between the ocean and the West Antarctic Ice Sheet. One mechanism that contributesto the cross-shelf transportis Ekmantransportinduced by along-slope currentsoverthe slope and shelf break. An investigation of this process is applied to the Amundsen Sea shelfbreak region, using recently acquiredandhistoricalfielddatatoguidetheanalyses.Along-slopecurrentswereobservedattransectsacross the eastern and western reaches of the Amundsen slope. Currents in the east flowed eastward, and currents farther west flowed westward. Under the eastward-flowing currents, hydrographic isolines sloped upward paralleling the seabed. In this layer, declining buoyancy forces rather than friction were bringing the velocity to zero at the seabed.Thebasin water in the easternpart ofthe shelfwas dominatedby wateroriginating from 800‐1000-m depth off shelf, suggesting that transport of such water across the shelf frequently occurs. The authors show that arrested Ekman layers mechanism can supply deep water to the shelf break in the eastern section, where it has access to the shelf. Because no unmodified off-shelf water was found on the shelf in the western part, bottom layer Ekman transport does not appear a likely mechanism for delivery of warm deep water to the western shelf area. Warming of the warm bottom water was most pronounced on the western shelf, where the deep-water temperature increased by 0.68C during the past decade.

Bacterial communities of surface mixed layer in the Pacific sector of the western Arctic Ocean during sea-ice melting.

From July to August 2010, the IBRV ARAON journeyed to the Pacific sector of the Arctic Ocean to monitor bacterial variation in Arctic summer surface-waters, and temperature, salinity, fluorescence, and nutrient concentrations were determined during the ice-melting season. Among the measured physicochemical parameters, we observed a strong negative correlation between temperature and salinity, and consequently hypothesized that the melting ice decreased water salinity. The bacterial community compositions of 15 samples, includicng seawater, sea-ice, and melting pond water, were determined using a pyrosequencing approach and were categorized into three habitats: (1) surface seawater, (2) ice core, and (3) melting pond. Analysis of these samples indicated the presence of local bacterial communities; a deduction that was further corroborated by the discovery of seawater- and ice-specific bacterial phylotypes. In all samples, the Alphaproteobacteria, Flavobacteria, and Gammaproteobacteria taxa composed the majority of the bacterial communities. Among these, Alphaproteobacteria was the most abundant and present in all samples, and its variation differed among the habitats studied. Linear regression analysis suggested that changes in salinity could affect the relative proportion of Alphaproteobacteria in the surface water. In addition, the species-sorting model was applied to evaluate the population dynamics and environmental heterogeneity in the bacterial communities of surface mixed layer in the Arctic Ocean during sea-ice melting.

Disintegration and acceleration of Thwaites Ice Shelf on the Amundsen Sea revealed from remote sensing measurements

Thwaites Ice Shelf in the Amundsen Sea is one of the biggest ice shelves in West Antarctica and is well known for significant mass changes. In the shear zone between Thwaites Glacier Tongue and its eastern ice shelf, shear stress forced by different flow rates of the ice shelves is causing the ice to break apart. A time series analysis of remote sensing data obtained by Landsat 7 Enhanced Thematic Mapper Plus (ETM+), TerraSAR-X, and airborne synthetic aperture radar (SAR) revealed that the shear zone has extended since 2006 and eventually disintegrated in 2008. We quantified the acceleration of Thwaites Ice Shelf with time by using the feature tracking method. The buttressing loss induced by the extension of the shear zone and progressive disintegration accelerated the flow of Thwaites Glacier Tongue, which in turn increased the shear stress on its eastern ice shelf. We determine causes of disintegration in the newly formed shear zone to be oceanic basal melting and structural weakening induced by Circumpolar Deep Water intrusion beneath the eastern ice shelf since 2000. The structural weakening was examined by using the density distribution of rifts and crevasses on the ice shelf, which were well identified from high-resolution SAR and optical satellite images.

Seasonal and interannual changes in the sound scattering layer at deep-sea floor in the Amundsen Sea, Antarctica

Vertical migration of zooplankton is ubiquitous behavior in marine plankton community; however, seasonal and interannual behavior are little observed in the deep sea under seasonal varying sea ice. Here, the first evidence that sound scattering layers of zooplankton can support the knowledge for understanding the effect of climate change is presented, based on four-years acoustic backscattering strengths in the Amundsen Sea, Antarctica. Amundsen Sea is a biological hotspot region with the rapid oceanic melting of the ice shelf as well as the most productive (per unit area) in the Antarctic. High-temporal resolution profiles of acoustic backscattering strength collected from a bottom-moored, upward looking Acoustic Doppler Current Profiler were examined to describe the temporal variation of sound scattering layers. Our observations show that sound scattering layers exhibited clear diel, seasonal, and interannual pattern associated with solar radiation, sea ice concentration, and phytoplankton biomass. The ...