Shoreline Morphology

This section describes the method and results of the digital elevation models at Pu‘ukoholā Heiau NHS and Kaloko-Honokōhau NHP.

Integrated Coastal Digital Elevation Model (C-DEM)

One objective of this study was to examine the vulnerability of the coastal area to overtopping, inundation and sea level change. An important component to these investigations is accurate detail of the topography at the coastline at various scales. C-DEMs were produced for Pu‘ukoholā Heiau NHS and Kaloko-Honokōhau NHP areas at three different grid sizes (1, 5, and 25 m) to examine wave setup and inundation scenarios. The 5-m grid size C-DEM was also used in the imagery orthorectification process.

Data Sources

Light Detection and Ranging (LiDAR) data were collected for the Federal Emergency Management Agency (FEMA) and the Army Corps of Engineers (ACE) by the Joint Airborne LiDAR Bathymetry Technical Center of Expertise (JALBTCX). FEMA terrain data extends landward from the water line to include the 15-m elevation contour at the time of collection. ACE SHOALS (Scanning Hydrographic Operational Airborne LiDAR System) data extends from just below the calm water surface to approximately 40 m (~130 ft) water depth or survey boundary offshore, whichever is reached first. Ten-meter USGS DEM data from spot elevations were used to fill areas that lie landward of the data bounds of these two data sources.

FEMA LiDAR data were received as two processed separate data products; bare earth returns and extracted features. The bare earth returns were used as the primary terrain dataset. Bare earth returns are a subset of the acquired LiDAR data, features in the landscape such as buildings, vegetation and structures are removed to leave ‘bare earth'. The extracted features from this process are retained in the ‘extracted features' dataset. We identified structures of interest for this study and replaced them into the final terrain LiDAR dataset. The data are vertically referenced to the Local Tidal Datum (LTD) which is usually a local iteration of the North American Vertical Datum of 1988 (NAVD88). SHOALS LiDAR was used as the submarine coverage for this study for the purpose of image orthorectification and wave inundation modeling. Data received were vertically referenced to local mean lower low water (MLLW) tidal elevation. USGS 10-m data were received as raster DEM files and converted to points using ArcMap toolbox function raster to point for each cell yielding a point file with 10 m horizontal spacing. The vertical reference of the USGS data is inferred to be MHW after Taylor et al. (2007). Coverage of elevation databases for each site is indicated in Figure 36.

Figure 36. Elevation source data extents for Pu‘ukoholā Heiau NHS (left) and Kaloko-Honokōhau NHP (right).  White space indicates No Data while the dark blue features were added into the final DEM from the extracted features product of the FEMA LiDAR datasets.

Figure 36. Elevation source data extents for Pu‘ukoholā Heiau NHS (left) and Kaloko-Honokōhau NHP (right). White space indicates No Data while the dark blue features were added into the final DEM from the extracted features product of the FEMA LiDAR datasets.

Methods – vertical datum migration

NAVD88 is specific to the continental US and does not exist for Hawai‘i. Survey data associated with the FEMA LiDAR indicates the vertical datum, which the data is referenced to, is an iteration of the LTD – based on the last (1975) leveling network – updated to the present 1983-01 tidal epoch (MSL) – based on the 3 Kawaihae tidal benchmarks (+0.16 m), and accounting for sea level rise between the epochs (-0.031 m). This superseding survey places FEMA LiDAR in a modernized MSL datum approximately 0.13 m above the Kawaihae Harbor MSL elevation. SHOALS data were received in the MLLW tidal datum based on a survey that regionally references the data to the closest tidal station.

The USGS 10-m DEM was used at Pu‘ukoholā Heiau NHS where some landward historical photo data required approximate elevations during the orthorectification process. No USGS DEM data were used at Kaloko-Honokōhau NHP since FEMA terrain LiDAR data exist for the entire region of interest.

All data were processed in ArcMap and vertical adjustments were made using the data field calculator for each dataset. Data gaps were not filled. SHOALS data were filtered to reject any elevation points above 0 and where there appears to be FEMA points on bare earth. This results in well characterized near-shore features. FEMA data appears to have been collected near a lower tide stage with inter-tidal features. Cultural features, such as fishpond walls and selected ground returns that were mis-classified as vegetation (algae is most cases), were re-introduced into the DEM. All masks for the terrain LiDAR data were manually digitized based on a 2-m resolution hillshade characterization of an interpolation of the point data to highlight the shore-water interface present in the data. The mask was used to define the shoreline boundary for both the terrain (seaward extend) and bathymetric (landward extent) data. Datasets were then spatially edited to remove overlap between the different elevation sources. The data merging process used ArcToolbox – Merge and a common elevation field between the databases. The result was an irregular point cloud including both on and off-shore elevation values. A natural neighbor interpolation within Arc 3D Analyst was chosen to create a raster DEM.

Results

FEMA LiDAR data were migrated from the modernized MSL to the local MLLW on the Kawaihae tide gauge by subtracting 0.415 m from the point elevation. This study found differences in the standard deviations of the overlapping bare earth coverage areas after this migration of 0.1 m at Pu‘ukoholā Heiau NHS and 0.22 m at Kaloko-Honokōhau NHP using all points within 0.5-m radius of the SHOALS data points. These values fall within the vertical accuracy associated with each data source and the control survey, which located the data within the vertical datum.

The results of merging of the databases created point clouds of more than 4.9 million points for Kaloko-Honokōhau NHP and more than 7.4 million for Pu‘ukoholā Heiau NHS. The results of the interpolation of the merged datasets were two 32-bit depth DEM grids for each area at 1 and 5-m horizontal resolution. The C-DEM was used in the orthorectification process and an XYZ format version of each grid was generated for modeling. Characterizations of the two generated 5 m C-DEMs are shown in Figure 37.

Figure 37. 0. 5 m horizontal resolution hillshade characterizations of the final C-DEMs for Pu‘ukoholā Heiau NHS (left) and Kaloko-Honokōhau NHP (right).  These DEMS are composites of SHOALS LiDAR (bathymetric), FEMA terrain LiDAR, and USGS 10 m DEM data.

Figure 37. 0. 5 m horizontal resolution hillshade characterizations of the final C-DEMs for Pu‘ukoholā Heiau NHS (left) and Kaloko-Honokōhau NHP (right). These DEMS are composites of SHOALS LiDAR (bathymetric), FEMA terrain LiDAR, and USGS 10 m DEM data.