COASTLINE
DATA - 1 ARC SECOND (ISCIENCES)
[Status: Not yet released.]
THE PROBLEM
Existing coastline datasets tend to be either too coarse in scale, have
serious registration differences when used with high resolution
raster datasets such as SRTM or Landsat data, or are incomplete.
THE SOLUTION
The ISciences Coastline Data is a global map
at 1 arc-second of resolution (approximately 30 meters
at the equator) defining which areas of the earth are
land and which are ocean or sea. We developed this dataset to provide a global land/sea mask that closely matches modern high-resolution raster datasets. The primary source is USGS SRTM
Water Body Data. ISciences Coastline Data has the advantage of high resolution, excellent co-registration, and is complete, without gaps due to data voids.
|
|
 |
Area of Interest: Several small islands in the Bungalaut Strait just south of the island of Siberut, West Sumatra, Indonesia.
Extent:1°49’44”S to 1°56’39”S; 99°13’16”E to 99°22’17”E |
|
|
THE STORY
The three images below help tell the story of how ISciences created an improved global coastlines map by starting with the best available data, then carefully correcting voids. Each image shows the same area of the Bungalaut Strait using overlays based on different data, and each is centered at 1° 52'N / 99° 17'E, with a scale of approximately 1:100,000, representing an area roughly 15km by 15km. Scroll to the larger images for complete details.
|
IMAGE 1:
 |
|
IMAGE 2:
 |
|
IMAGE 3:
 |
|
|
| IMAGE 1: Coastline Composite (below) displays data from three coastline datasets: VMAPO (Digital Chart of the World-NIMA) in cyan, World Vector Shoreline (NOAA-NGA) in red, and ISciences Coastline 1 Arc Second (ISciences) in yellow. These data are overlaid on an image from NaturalVue 2000™, a natural color, Landsat-7 derived image circa year 2000 with a resolution of .5 arc sec. NaturalVue 2000™ is a trademarked product available from MDA Federal in Rockville, MD (NaturalVue © MDA Federal Inc, 2008, www.mdafederal.com). The NaturalVue 2000™ image provides a realistic basis for comparison. Note that the ISciences coastline vector in yellow closely matches the NaturalVue 2000™ image, while the other two coastline vectors in red and cyan do not. Also note the increasing level of detail in the three vectors: Cyan (VMAPO) 1:1,000,000); red (World Vector Shoreline) 1:250,000; and yellow (ISciences) 1: 100,000. |
|
| |
IMAGE 2: SRTM Water Body Data (below) To create the ISciences Coastline dataset we began with the SRTM Water Body Data, visible as a violet outline against the NaturalVue 2000™ image below (NaturalVue © MDA Federal Inc, 2008, www.mdafederal.com). Note that the SRTM Water Body Data follow the island shorelines in some parts of the image and deviate considerably in others - the large areas to the right of the islands. These large areas that don't match the shoreline represent data voids.
The SRTM Water Body Data files are a by-product of the data editing performed by the National Geospatial-Intelligence Agency (NGA) to produce the finished SRTM Digital Terrain Elevation Data Level 2 (DTED® 2). In accordance with the DTED® 2 specification, the terrain elevation data were edited to portray water bodies that meet minimum capture criteria. Ocean, lake and river shorelines were identified and delineated. Lake elevations were set to a constant value. Ocean elevations were set to zero. After this processing was done, the shorelines from the one arc second DTED® 2 were saved as shapefiles.
The key factor that prevents ocean polygons from being an acceptable coastline definition is that voids in the SRTM data were not filled. Thus, the
Water Body dataset is delineating a coastline in some instances, and
in other areas is tracing the limits of a void. |
|
| |
| IMAGE 3: Corrected SRTM (below) displays data voids in the SRTM Water Body Data and displays the corrections ISciences made to the voided areas to more accurately represent shoreline. To create a corrected shoreline ISciences
located the areas where the Water Body dataset was
delineating a void
and corrected it to the actual shoreline based on several other
datasets (JERS images, NOAA's World Vector Shoreline, NGA Global Shoreline Data,
Landsat). Over 350 of the 3,700 1-degree tiles that define
the shoreline within the SRTM coverage area (+/- 60
degrees latitude)
were edited during this process. The same sources used to correct
for voids were also used to replace missing
tiles within the SRTM coverage area, and for
all of the areas outside of outside of the SRTM coverage area. The resulting
coastline map is 1,296,000 pixels wide by 648,000 pixels
high, an image of almost 850 billion pixels. The ISciences Coastline dataset has the advantages of high resolution (1:100,000), excellent co-registration with SRTM and Landsat, and is complete, without gaps or spurious features due to voids. |
|
| NOTE: ISciences' Coastline Data is used to produce administrative boundaries for the Political Boundaries Maps in the TerraViva! collection. |
|
