Courtney Reents – Measuring the Extent of Oceanic Tidal and Saline Influence in the Coastal Marshland of the Ka’a’awa Valley

Abstract

The purpose of this research is to determine the extent of oceanic saline and tidal influence in the coastal marsh region at the eastern edge of Ka’a’awa Valley. Among the data assembled to investigate the area are discrete point conductivity and temperature readings; one and two-day Levelogger measurements of temperature, conductivity and water-level; existing LiDAR, soil and wetland data; in-person surveying of present vegetation; and visual overhead WorldView 2 8-band imagery. The discrete conductivity and temperature readings are mapped as two separate polygons based on geographic location, and then interpolated individually to produce raster images, which are then overlaid with various other layers to highlight trends or correlations. The levelogger data are graphed and similarly assessed for trends. Final assimilation of all data shows that the extent of oceanic influence is primarily concentrated in the eastern-most area, just beside the coast, and that any salinity trends in the more inland area are likely the result of anthropogenic agricultural influence or water temperature affecting conductivity readings.

Introduction

The Ka’a’awa Valley is located on the eastern side of the island of Oahu, in Hawai’i. It is the property of Kualoa Ranch, a working cattle ranch and, as a result of its natural beauty and relatively undeveloped land, a filming location for many major motion pictures and television shows. As a result, the valley is something of a tourist attraction, with daily ATV, horse, bicycle and bus tours and an infrastructure of roads and trails to support them. Additionally, a large percentage of the land is dedicated to cattle pasturing as well as a hydroponics endeavor on the eastern edge that, with further investigation, could have the potential to be of consequence to this study, but is not addressed directly in this paper.

Given its orientation, the valley receives ample rainfall yearly and has a notable hydrological network that reaches its ultimatum in the main stream channel, which runs through the center of the valley before turning north and reaching the ocean. The area of study for this investigation is not, at least superficially, connected to this main channel, but instead lies to the south-east of it, delineated on the northwest by an abrupt change in elevation of several meters. It is comprised of two almost certainly man-made ponds, one potentially natural pond, several marshy interim areas and ditches, and, closer to the coast opposite an impenetrable Hau tree (Hibiscus tiliaceus) stand, a tidally influenced saltwater marsh dominated by a mangrove (Rhizophora mangle) stand. (see figure 1)

The aim of this research is to determine to what extent the ocean influences this marsh area in terms of salinity and tide through the use of empirical measurements of conductivity (as an analog for salinity), temperature and water level throughout the previously defined area. Outside the scope of this project, but an area of potential further research would be an elaboration and exploration of any anthropogenic influence that may be occurring in this region as a result of the agricultural endeavors of the ranch.

id=”attachment_4898″ align=”aligncenter” width=”671″ caption=”Figure 1 – area of study”[1]

Instruments, Software and Methodology

The instruments used in the collection of this data were an eTrex Garmin GPS, an Oakton Conductivity and Temperature Meter and two Solinst Leveloggers. For preliminary planning, Google Earth and 2011 WorldView 2 imagery were used to visually identify regions of interest, and opposite this, for data processing, ESRI ArcMap and Excel were utilized to manage and process the data and to create visual representations.

The first step in the data collection process was to define the “what” and the “where”. Given the straightforward nature of the study, it was a simple decision to define conductivity, temperature and water level as the attributes of interest. Areas of interest were decided based on a first-day in-person survey of the area to identify potential phenomena and accessibility problems and to define an informal hierarchy of importance and the general extent of the total marsh region. Due to an impenetrably dense vegetation growth near the center of the study area, the data were collected in two distinct regions instead of a continuous area: one a saltwater marsh close to the shore with mangroves and standing water, and the other a series of ponds and marshes farther inland. For the purposes of this paper, the coastal area will be referred to as the “coastal marsh” while the more inland area will be referred to as the “inland marsh”. (see figure 2)

Three days were spent collecting the actual discrete point data in the form of lines of points for linear water features, and outlined polygons for ponds or poorly accessible marsh water features. For each point measured, a GPS value was taken and all data was manually recorded as a unit in a field notebook for later computer entry. Because of this method, the GPS points are not the exact location of the point measurements, but still accurately reflect spatial trends in the measurements taken.

These methods worked well for the inland marsh, but were insufficient when applied to the coastal marsh because the heavy presence of saltwater in this area caused conductivity to change drastically according to water depth. To account for this, two measurements were taken for each point of adequate depth: one marked “deep” and the other “shallow”. The area of this nature was not necessarily widespread enough to produce an individual map, except to show the extent of the occurrence, but a video was also recorded to demonstrate the phenomenon for later viewing.

Two leveloggers were placed in open-water areas, one in a slightly saline ditch feature, and one in a marshier saline feature, 0.15 miles away from the coast in the inland marsh area for two and one days, respectively. This time difference was due to timing constraints, but did not appear to have any effect on the accuracy of the data, as both were in place for at least two tide cycles. The loggers were set to measure every fifteen minutes, which provided a high enough temporal resolution to account for tidal changes. (see figure 3)

id=”attachment_4887″ align=”aligncenter” width=”496″ caption=”Figure 2 – Inland marsh in green and coastal marsh in blue”[2]

id=”attachment_4892″ align=”aligncenter” width=”620″ caption=”Figure 3 – Locations of the two Leveloggers”[3]

Data Analysis

The initial data from the collection came in two basic forms, both underlain by raw data in attribute tables: maps of conductivity and temperature by location, and graphs of conductivity and temperature over time. In their most rudimentary form, the maps were a series of points, color-coded to indicate conductivity and temperature and later marked as un-outlined squares to improve trend visibility. These points were then separated into two separate sections – the coastal marsh and the inland marsh – and then, after some experimentation with the sections together and apart, were run as two individual regions through an inverse distance weighting interpolation in ESRI ArcMap 10.1. Inverse distance weighting was chosen as the interpolation method because it operates on a local scale and does not change the values of the points it runs through, which is important for maintaining the range and trends of the data. The settings for these interpolations included a basic polygon mask to define the sections, a variable search radius, and a point number of eight. The scales were then lined up so that the polygons could be visually and simultaneously compared. In addition, the points were overlain on wetlands and soils data from the State of Hawaii and the US Fish and Wildlife Service, as well as a processed LiDAR image to seek out elevation trends.

The graphs were created in Excel from the levelogger data, and included water level, temperature and conductivity over time, as well as a graph using tidal cycle information obtained online for comparison. Beyond this, the graphs did not require much further manipulation or processing, except to line them up together to check for trends. In addition, one extra graph was made from the discrete point data that mapped both conductivity and temperature by point number, in order to verify the existence of a temperature/conductivity trend previously found in the levelogger data.

Findings

The results of the discrete points were fairly conclusive as to the extent of saltwater and freshwater throughout the marshes. In the coastal marsh area, the water did not quite connect to the ocean directly, but later data collection by another group (Gordan Buckingham) showed high levels of salt water running through the sand, at the same level of conductivity later detectable farther into the marsh. This saltwater was highly concentrated on the bottom of the marsh water (32.00 milliSiemens/cm), and the conductivity reading dropped steadily as the depth of the meter decreased (as low as 5.00 mS/cm on the top). By the back of the coastal marsh, the water depth was much lower and a deep and shallow reading were no longer possible. Individual readings marked the water at 2.00 milliSiemens/cm or less (drinking water being around 0.40 mS/cm and ocean water being around 58.00 mS/cm). (see figure 4)

Soil data later overlain marked this area as being well-drained and highly permeable, however the LiDAR data demonstrated that it was also sitting below sea level, which may contribute to the reason for the existence of standing water. (see figure 5) The inland marsh was partially defined as marsh soil, and partially defined as different clays. The clays varied in drainage and permeability, but the marsh area, naturally, was listed as very poorly drained with frequent flooding. This marsh area fell where the Hau trees prevented accessibility, and offered some insight as to what is happening in the area that as of yet is not visible by any means. An additional wetland area overlay closely matched this trend and included the ponds and the coastal marsh as wetlands as well. (see figure 6)

The levelogger data proved relatively insignificant as far as demonstrating trends, but did help to solidify the study of oceanic extent by evidencing a lack of tidal influence in the inland marsh. The water level did vary significantly between each individual measurement, but on a large scale, there was only slight variation with little particular pattern. (see figures 7 and 8). Temperature and conductivity both changed simultaneously with day and night, which highlighted a positive correlation between these two measurements that proved useful in explaining later data. (see figures 9, 10 and 11)

In the inland marsh, strange conductivity patterns emerged in a small stream feeding the area with slightly saline water, which was distributed unevenly throughout the rest of the ponds. The farthest north pond was the most conductive, at about 1.00 milliSiemens/centimeter, while the farthest south was the least, at about 0.450 mS/cm. The center pond registered at 0.550 mS/cm, and the surrounding marshy ditches at around 0.780 mS/cm. It was in one of these ditches that the first levelogger was placed. South of the freshwater pond, there was also a marshy area with more saline water at around 0.850 mS/cm, also the location of the second levelogger.

The temperature data showed very few trends, except increased temperature in standing water, decreased temperature in flowing water, and extremely decreased temperatures in the coastal marsh, likely due to the dense canopy overhead. (see figure 12)

id=”attachment_4891″ align=”aligncenter” width=”300″ caption=”Figure 4 – Interpolation of conductivity data using inverse distance weighting”[4]

id=”attachment_4893″ align=”aligncenter” width=”300″ caption=”Figure 5 – Conductivity points overlain on LiDAR elevation data”[5]

id=”attachment_4889″ align=”aligncenter” width=”300″ caption=”Figure 6 – Conductivity points overlain on soil polygons”[6]

id=”attachment_4902″ align=”aligncenter” width=”621″ caption=”Figure 7 – Water level change over a two-day period (Levelogger 1)”[7]

id=”attachment_4901″ align=”aligncenter” width=”570″ caption=”Figure 8 – Tide level recorded for nearby beach on corresponding dates”[8]

id=”attachment_4888″ align=”aligncenter” width=”535″ caption=”Figure 9 – Water conductivity over two-day period (Levelogger 1)”[9]

id=”attachment_4900″ align=”aligncenter” width=”512″ caption=”Figure 10 – Water temperature over a two-day period (Levelogger 1)”[10]

id=”attachment_4896″ align=”aligncenter” width=”1024″ caption=”Figure 11 – Temperature and conductivity graphed against point number – to demonstrate correlation between higher temperature and higher conductivity reading”[11]

id=”attachment_4899″ align=”aligncenter” width=”300″ caption=”Figure 12 – temperature measurements displayed as points”[12]

Conclusion

After analyzing and comparing the collected data, the conclusion, or rather, the revised hypothesis, is that the extent of oceanic tidal and saline influence ends around the farthest west extent of the coastal marsh, at the eastern extent of the Hau stand. This is the point at which the conductivity of the water nearly equals that of the water on the other side of the stand, and is near the point where the soil changes from loam to marsh and, consequently, well to poorly drained. It appears that the saltwater comes in from the ocean, either through the sand or through the ground water, and sits in the coastal swamp, keeping the soil saline. It stays near the coastline, meaning that the salt content of the water fades as one moves deeper inland, and doesn’t make it past the marsh soil region at all. This is likely due to the outflowing freshwater coming through this area, where a unidirectional flow prevents the saltwater from moving any farther into the soil or surface water. This may or may not be related to the soil boundary at this location, and would warrant further research to see the exact spatial arrangement of these boundaries.

In further support of this point, some basic vegetation research and identification, which defines the mangroves as Rhizophora mangle and the Haus as Hibiscus tiliaceus, supports the statements of soil salinity. According to Norman C. Duke, James A. Allen, Craig R. Elevitch and Lex A.J. Thomson of the Traditional Tree Initiative, the Atlantic-East Pacific red mangrove grows best in saline soils, between 12.5 and 40.6 mS/cm, while the beach hibiscus (universal name for Hau) only tolerates them. The change between these two types of vegetation matches almost exactly to the edge of measured salinity in the coastal marsh, and may indicate that beyond this point, the soil is less saline than closer to the coast. (see figure 13)

Finally, as there was no tidal influence found inland, it is highly likely that any inland salinity trends are due to anthropogenic causes or natural soil salinity, and are not a result of general oceanic influence. In addition, the correlation visible between temperature and conductivity reading could explain, at least for the northernmost pond, the apparently higher levels of salinity found there, though this trend does not seem to continue throughout the remaining inland ponds and marshes. Further investigation, as well as input from the land managers, would be required to differentiate between natural effects, measurement error, and agricultural influence.

id=”attachment_4895″ align=”aligncenter” width=”300″ caption=”Figure 13 – Informal classification of mangrove and Hau trees in coastal marsh and intermittent area”[13]

Limitations

The most notable limitation that affected data collection was that of accessibility. Due to a variety of factors – vegetation, water, safety concerns, etc. – certain parts of the region of study are lacking in data. By far the most compromising obstacles were the Hau trees, which grew in an extremely dense stand between the coastal marsh and the inland marsh, thereby separating them in terms of data. (see figure 14) The tangled and close-growing nature of the stand prevented both ground-level entry and overhead remote sensing to uncover what was occurring on the ground below; even the LiDAR data mistakenly displays a mound of around nine meters in elevation, which better reflects the canopy height than any true underlying topography. (see figure 5) Given the time allotted, the only way to compensate for this was to assess the data on either side of the area and make an educated guess of what lay within, which ultimately sufficed for this project.

In terms of water and safety, the inaccessibility problem became more a product of gear and manpower. Given the unknown depth and quality of the water, it was inadvisable to enter any ponds or marshes without proper wading clothes, which were not available at the moment. Safety was also a concern, as a lack of people meant that if a solo researcher ended up in a dangerous situation, there was no one else there to help.

The final issue came as a result of GPS network limitation, in that a dense overhead forest canopy in some areas caused a lack of satellite reception, which manifested in the form of stray points on the map, or no ability to measure a point at all. The correction for this, if unorthodox, was to reach a higher elevation by climbing a nearby tree to get a better signal and achieve a more accurate latitude and longitude (due to constant, overwhelming error, the elevation as marked by the GPS unit was never recorded, so there was no issue here).

Despite these limitations, as well as time constraints, the data collection process was made as thorough as possible with the resources available. Ultimately, the analysis functioned well around these problem areas, and the issues that did arise were mitigated in their negative effect on the final product.

id=”attachment_4890″ align=”aligncenter” width=”225″ caption=”Figure 14 – example of dense Hau branches, which restrict accessibility”[14]

Bibliography:

Duke, Norman C., and James A. Allen. Rhizophora Mangle, R. Samoensis, R. Racemosa, R. X Harrisonii. Publication. US Department of Agriculture, Apr. 2006. Web. 27 June 2013.

Elevitch, Craig R., and Lex AJ Thomson. Hibiscus Tiliaceus (beach Hibiscus). Publication. US Department of Agriculture, Apr. 2006. Web. 27 June 2013.

HI_Wetlands_Metadata. N.p.: US Fish and Wildlife Service, 2 June 2010. SHP.

“Monthly Tides for Moku O Loe, Hawaii.” Www.hawaiitides.com. N.p., n.d. Web.

Oah_streets. Honolulu: City and County of Honolulu, 11 Dec. 2011. SHP.

OaSoil. N.p.: SSURGO Data, 1972. SHP.