Saturday, January 25, 2014

Importing bathymetry and coastline data in R

After noticing some frustrating inaccuracies with the high-resolution world coastlines and national boundaries database found in worldHires from the package mapdata (based on CIA World Data Bank II data), I decided to look into other options. Although listed as "depreciated", the data found in NOAAs online "Coastline Extractor" is a big step forward. There seem to be more up-to-date products, but this served my needs for the moment, and I thought I'd pass along the address to other R users. I exported the data in ASCII "Matlab" format, which is basically just a 2-column text file (.dat) with NaN's in the rows that separate line segments.

I've also discovered the bathymetry / topography data from GEBCO. Again, very easy to import into R from the netCDF files.

The above map of the Galapagos Archipelago illustrates the quality of both datasets. It also shows the comparison of coastline accuracy between World Vector Shoreline (1:250,000), world (map package), and worldHires (mapdata package) datasets. Obviously, the low-resolution world data only makes sense for quick plotting at large scales, but the high-resolution data is as much as 1/10° off in some locations. I noticed these errors for the first time when trying to map some data for smaller coastal bays. It drove me crazy trying to figure out where the errors were - in my data locations or the map itself. Bathymetry used in the map was 30 arc-second resolution GEBCO data.

[EDIT: The comparison of coastline data now includes the high resolution data from the rworldmap package.]

A more detailed description export settings:
  • Coastline data (from 'Coastline Extractor') :
    • Coastline database: World Vector Shoreline (1:250,000)
    • Compression method for extracted ASCII data: None
    • Coast Format options: Matlab
    • Coast Preview options: GMT Plot
  • Bathymetry / topography data [link]:
    • General Bathymetric Chart of the Oceans (GEBCO) : GEBCO_08 Grid (30 arc-second resolution)
Here's another example of bathymetry / topography data for the western Pacific (1 minute resolution GEBCO data):

For both maps, I took inspiration for the color palettes from GMT. The rgb color levels of these palettes have got to be documented somwhere, but I gave up looking after a while and managed to hack their levels from color scales contained in .png files [link].

Below is the R code to reproduce the figures.

GMT standard color palettes

GMT (Generic Mapping Tools) ( is a great mapping tool. I'm hoping to use it more in the future, but for the meantime I wanted to recreate some of the it's standard color palettes in R. Unfortunately, I couldn't find documentation of the precise rgb color levels used, so I ended up "stealing" them from the .png images on this website:

Here's the result:

Here's how I extracted the color levels from the .png images: