Skip to Local Navigation
Skip to Content
California State University, Long Beach
CSULB Geospatial Research and Mapping (GRAM) Field Program
Print this pageAdd this page to your favoritesSelect a font sizeSelect a small fontSelect a medium fontSelect a large font
 

Thomas Hervey – Investigation of Offshore Submarine Groundwater Discharge (SGD) and the Relationship Between Fresh Water and Coral Communities in Ka’a'awa, HI

Contents

[ hide ]

    Thomas Hervey
    University of Maryland Baltimore County


    Investigation of Offshore Submarine Groundwater Discharge (SGD) and the Relationship Between Fresh Water and Coral Communities in Ka’a'awa, HI

    NSF Grant #: 1005258

    Introduction

    The Ka’a'awa Valley is a 1742 acre valley that receives 1700mm (Giambelluca) of rainfall annually accumulating over 1 million m³  monthy. Width a few swamps and only one riparian corridor that discharges less than 11% total accumulation, there is an obviously large amount of evapotransportation and groundwater collection. Submarine groundwater discharge (SGD), also known as “seeps,” is a commonly studied form of coastal freshwater percolation. While there are known SGD locations along the neighboring Koaloa and Kahana Bay coasts, little mapping and data collection has occurred in front of the Ka’a'awa valley.


    Coral communities are are an delicate foundation for Oahu shallow water ecosystems. Providing both a habitat and a food source for fishes, algae and other marine life, coral is essential for a healthy environment. Since Hawaiian coral lives in a very specific environment, shifts and variations in water composition can deem areas uninhabitable for coral.


    The purpose if this project was to address the question, “can coastal SGD locations be identified and can a correlation between freshwater and coral community locations be determined?” Broken up into two main sections, this project focused on measuring conductivity and temperature to locate percent seawater variations and then observing changes in bathymetry and ocean floor surfaces to visually analyze if there is a coral location relationship.


    Before conducting field observations, it was necessary to gain an understanding about SGD characteristics, negative environmental factors for coral, and the spectral reflectance of coral for imagery analysis.

     

    Figure 1. Schematic depiction (no scale) of processes associated with SGD. Arrows indicate fluid movement (Taniguchi)

    It can often be difficult to find how far off shore submarine groundwater discharge occurs. As noted in Figure 1., the difference between confined and unconfined aquifers can yield both different amounts of discharge at different distances, but mapping the water table can give a good clue as to SGD locations. When capstone isn’t present like on the Ka’a'awa shore, groundwater isn’t as suppressed making SGDs more likely. However, based on soil types, slope, vegetation, rainfall and several other factors, the amount and expanse of seep water can vary substantially. Figure 2 from a University of Hawaii poster demonstrates how surface temperature can vary substantially more than 200m offshore.


     Figure 2. Sea surface temperature (SST) map produced from August 2005 aerial TIR survey over coastal waters (Johnson)

    Hawaiian coral reefs are most commonly found at temperatures between 25-31°C and a salinity range of 34-37ppt (~55,000 microsiemens). In addition, reefs need sufficient light to, “maintain symbiotic association between corals and the symbiotic algae” (Florida Museum of Natural History). With heavy silt and sedimentation or deeply cut channels, lack of light inhibits photosynthesis by the symbiotic algae. For this project, several indicators including lowered light level, high silt and sediment cover, and the lack of symbiotic algae were noted as detrimental to coral habitats.


    In order to work with remote sensing imagery, it was necessary to understand the spectral response of different ocean floor surfaces. To discern the difference between sand, rock and coral, readings suggest that there is a, “very strong spectral reflectance from about 700nm towards the longer wavelength range” (Miyazaki). Other readings suggested that, “reflectance spectra between 650-690nm is dependent on chlorophyll-a concentration and can be sued to discriminate bare send with no algal component from chlorophyll-a containing benthos” (Headly). These ranges would be taken into account when trying to classify aerial imagery.


     

    Methodology

    1. Submarine groundwater discharge (SGD) spots were located using Solinst LevelLogger 3001s to record conductivity and temperature measurements. Colder water with lower salinity were the strongest indicators of fresh water. To monitor location, a Garmin Etrex Venture HC was kept with each LevelLogger. With a LevelLogger on two kayaks, systematic parallel passes with varying offshore distances would collect water samples every 30 seconds (day 1) or every 10 seconds (day 2). Collection starting times ranged from 10am – 12pm.
    2. Conductivity and temperature data were post-processed by syncing the devices’ timestamps, joining measurement data with longitude and latitude coordinates. Using control seawater and freshwater samples, the LevelLoggers were calibrated to the control for easier visualization analysis.
    3. For each day (June 19th and June 21st), conductivity was plotted against temperature to observe if there was a correlation. Once a sufficient positive correlation was observed, coordinate data was gathered for the points that had both low temperature and conductivity. These locations were then revisited for further freshwater analysis. Besides a few outliers, the data highlighted on figure 3a, 3b and 3c all correspond to the same rough location.

    Figure 3a,b. Correlation between conductivity and temperature of the effective LevelLogger for each collection day

    1. To begin coral community location analysis, geotagged photos of coral changes were taken approximately 30-40 meters off shore along the entire valley. This assisted by ground truthing both WorldView 2 and Soest LiDAR shoals bathymetry data. Images were taken when there was a strong homogenous surface (ex. only sand in sight), or at locations with defined surface change (ex. transition from sand to low coral).
    2. Using Erdas Imagine 2013, a supervised classification was generated from 2011 WorldView 2 imagery of the Ka’a'awa valley coastal region. After pansharpening the imagery in order to work with a higher (0.5m) resolution, the spectral principal component analysis tool was run, yielding uncorrelated data to highlight what’s not obvious in false color imagery. Based on readings, several band combinations (figure 4a.: 4,3,1  figure 4b.: 7,4,2) were manipulated to discern sand, rock and coral variations. Using the geotagged coral images as a reference, a supervised classification was then generated from several signature editor samples including sand, rock, mixed sand, low coral, medium coral, high coral, deep ocean, and wave.

    Figure 4a,b. Erdas multispectral band classifications to highlight differences between submarine surfaces

    1. Using LiDAR shoals bathymetry courtesy of Soest, an Inverse Distance Weighting (IDW) interpolation was created in ArcMap (figure 5). This was useful for understanding depth and the likely change in light penetration.
    2. Once conductivity, temperature, surface classification, bathymetry and geotagged photograph data were organized and overlaid, analysis began by visually observing trends in freshwater and coral community locations. With WorldView2 true color imagery as a base, comparisons were made between depth and surface coverage with remotely sensed and ground truthed coral imagery. Due to issues with the LevelLogger 3001s, limited analysis could be done on freshwater and SGD locations.

     

    Results

                Figure 5 represents a compilation of three different days of data collection out on kayaks. Figure 6 represents a zoomed in selection of figure 5 containing our main study area.

     




    Figure 5. Conductivity compilation

    Figure 6. Focused conductivity compilation

    With outliers removed, conductivity ranged from 42,235 – 57840 microsiemens. Lower conductivity readings near the shore suggested a lower salinity and thus higher percentage of freshwater. Interest points 1 and 2 show the lowest saltwater concentrations within the study. However, based on the red lines where surface streams reach the coast, it’s likely that these readings may be heavily influenced by surface water. Areas further out suggested a quick diffusion of freshwater since samples quickly jump to 55,000-56,000. Interest point 3 highlights the continued data collected on 6/19/13. However, based on figure 5, the rest of the data showed high conductivity with little variation, suggesting steady salt water. Interest point 4 highlights two lines of data collected on different days. These different days appeared unusual but can be explained by tidal changes. Based on these figures, there was very little change in conductivity across the entire Ka’a'awa coast. This could have been because the samples were too small, the SGDs are farther out than expected, or the SGDs don’t produce enough water to prevent rapid dilution.

    Figure 7 represents temperature samples collected in parallel with conductivity from the LevelLoggers. In addition to variations between days, there were between hours of the same day. This led to the data being mostly unimportant. However, the two locations on either side of the point in the middle of the valley did show a slight decrease in temperature ranging from 24.9 – 25.5°c from 0 – 20 meters off the shore. Interest point 1 highlights a colder water farther off shore than warmer water. This could suggest either an SGD farther from shore and/or a poor habitat for coral. When compared to surface, there. Interest point 2 shows the variation in temperature based on time of day at the launch location. Interest point 3 highlights the highest temperature area which was likely a warm water eddy.

    Figure 7. Focused temperature compilation

    Figure 8 represents the resulting classification of submarine surfaces. Similar to the unclassified WorldView2 imagery, surfaces like rock sediment and sand are easily recognized.

    Figure 8. Submarine surface classification

    1)

    2)

     

     

     

     

    3)

     

    The three interest points above are example locations where ground truthed images match the classification. These images show the difference in habitats for coral. For example, interest point 1 is both deep based on the bathymetry in figure 9 and covered in a rocky silt covered surface, which isn’t a good coral habitat. However, interest point 2 shows tall coral covered in algae with sufficient light and little covering sediment.

    Figure 9. 1 meter bathymetery

    Figure 9 shows a bathymetry 1 meter IDW interpolation useful for estimating both light and silt coverage. Matching surface classification, it’s obvious that coral remains at, and is likely the cause of a higher elevation. Highlighted are current surface water discharge locations as well as possible SGD locations. While the SGD locations aren’t likely spots, they show characteristics similar to the close by surface discharge areas including lower elevation, sandier and rockier surface, less coral coverage, and slightly overall temperatures and conductivities.

     

     

    Limitations

                Due to the nature of hydrological data collection, there were a handful setup, user, processing errors. The largest issue with the project was the necessity to remove one of the LevelLogger’s data for two days of kayaking. From calibration and/or drifting issues, much of the 874 LevelLogger’s data had both inverse and inconsistent results. In addition, due to user error and time constraints, calibration efforts for the 874 LevelLogger were not successfully completed. The LevelLogger samples were also likely not large, long or diverse enough to accurately locate SGD locations.

    Besides calibration issues, sonar, GPR and thermal technologies weren’t utilized. These tools would have been helpful to respectively reinforce bathymetry & depth data, water table changes, and cold water plume locations. The figures below are example images of these technologies that could be used in future research.

    Figure a. (top)Starfish Scanline sonar imagery attached to a kayak for Scott Honda’s project

    Figure b. (right) Dr. Becker testing the GSSI 200mHZ GPR

    Figure c. (bottom) FLIR Tau 640 thermal imagery; circle references person size

     

     

    Conclusion & Future Work 

                As a result of this project, I was unable to effectively locate SGD locations along the Ka’a’awa valley coast. With a narrow collection of temperature and conductivity data, the areas with lower observed salinity seemed to quickly difuse and have little lasting evidence. Based on a lowered conductivity, deeper submarine surfaces and the lack of heavy coral composition, highlighted ares in figure 10 could be possible locations of SGDs. However, based on limited data, it is more likely that these areas are a result of currents moving surface water drainage southward. More concrete evidence would have pointed to a lower conductivity around 25,000 – 30,000 microsiemens.

    Even without diverse temperature and conductivity data, there appeared to be a correlation between surface silt and depth and coral community location. As areas became shallower, there was more light penetration. In addition, shallower surfaces usually had less silt often from active moving water and crashing waves. The WorldView2 imagery used to classify submarine surfaces and the Soest LiDAR data used to calculate bathymetry were both useful. Since ground truthed snorkel images revealed similar patterns, the results support introductory findings that coral communities live better in shallower waters with less surface silt and little variation away from 25-31°C and 55,000 microsiemens. This supports preliminary readings research information.

    Since there were a lot of project limitations and time restrains, it would be effective to continue this project and recollect some portions of data. For future work, I would like to gather (well calibrated) conductivity and temperature data again over a longer period of time and over a larger study area. This should give more evidence of SGD locations. I would also like to explore the use of GPR to map fluctuation the water table, use thermal imagery to compare with LevelLogger 3001 temperature measurements, and use Sonar to compare with LiDAR bathymetery data.

     

     

    Acknowledgements

                The Summer 2013 NSF-REU GRAM research experience hosted by California State University Long Beach has been one of the most invigorating and influential academic trips that I have been on. With the assistance from university faculty and graduate assistants, as well as Oahu locals, I’ve gained an incredible amount of fieldwork insight. Through the process of observation, data collection, collaboration, and analysis, I’ve been able to familiarize myself with the fieldwork research environment and understand the nature of its challenges, requirements and iterative processes.

    I want to give thanks to all of the professors involved especially Dr. Becker, Dr. Liop, Dr. Wechsler, Dr. Lee, and Dr. Hunt. I want to give a special thanks to all of our teaching assistants including Briton Voorhees, Paul Nesbit, Scott Winslow, Mike Ferris, Emily Allen, Michelle Baroldi for both their organizational and academic assistance. In addition, this trip would not have been possible without Ted Ralston, John Morgan and David Morgan as local Oahu trip coordinators.

    This material is based upon work supported by the National Science Foundation under Grant No. 1005258.

     

     

     

     

    References

    Giambelluca, T.W., Q. Chen, A.G. Frazier, J.P. Price, Y.-L. Chen, P.-S. Chu, J.K. Eischeid, and D.M. Delparte, 2013: Online Rainfall Atlas of Hawai‘i. Bull. Amer. Meteor. Soc. 94, 313-316, doi: 10.1175/BAMS-D-11-00228.1.

     

    Headley, J.; Mumby, P., “Biological and remote sensing perspectives of pigmentation in coral reef organisms,” Advances in Marine Biology

    URL: http://www.marinespatialecologylab.org/wp-content/uploads/2010/11/Hedley-and-Mumby-2002.pdf

     

    Johnson, A.; “Aerial infrared imaging reveals nutrient-rich groundwater inputs to the ocean,” American Geophysical Union, 2008. Geophysical Research Letters, vol. 35, no.15, 16 Aug 2008;

     

    Karen E. Joyce and Stuart R. Phinn
    “Spectral index development for mapping live coral cover,” J. Appl. Remote Sens. 7(1), 073590 (Feb 05, 2013). ; http://dx.doi.org/10.1117/1.JRS.7.073590

     

     

    Miyazaki, T.; Harashima, A., “Measuring the spectral signatures of coral reefs,” Geoscience and Remote Sensing Symposium, 1993. IGARSS ’93. Better Understanding of Earth Environment., International , vol., no., pp.693,695 vol.2, 18-21 Aug 1993
    doi: 10.1109/IGARSS.1993.322240

    URL: http://ieeexplore.ieee.org/stamp/stamp.jsptp=&arnumber=322240&isnumber=7705

     

    Taniguchi, M.; Burnett, W.;  Cable J.; Turner, J. “Investigation of submarine groundwater discharge,” HYDROLOGICAL PROCESSES; Hydrol. Process. 16, 2115–2129 (2002); Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hyp.1145

     

    “The Benefits of the 8 Spectral Bands of WorldView-2”  DigitalGlobe, White Paper 2010.

    URL: http://www.digitalglobe.com/downloads/WorldView-2_8-Band_Applications_Whitepaper.pdf

     

    http://www.flmnh.ufl.edu/fish/southflorida/coral/habitat.html

     

      Important Dates

      Field Locations