Takes in a raster dataset, changes the values to make it binary (i.e. 1/0) and then out pops a shapefile which adopts the values of the input raster. Works on linux although not tested extensively.
If we have a raster with some values (say 0, 1 and 2 perhaps representing ocean, land and ice) like this:
We can reclassify the raster to 2 values (e.g. 0 and 1 perhaps representing ocean and everything else) and then convert the raster to a vector getting this:
The mask passed in is a transformed version of the GIMP DEM as modified following Morlighem et al. 2014. A 5km post version of the mask is available in ./raster and the R code is set up to use this (i.e. the paths are hard-wired).
For immediate use, a 100m resolution vector outline is available in ./shp
However, the code framework should work for any mask, just change the path and the values which you need to alter to make it 1/0.
The code is taken largely from this post: https://johnbaumgartner.wordpress.com/2012/07/26/getting-rasters-into-shape-from-r/
Linking the R and python libraries on windows is less easily implemented. So, if running on windows, have a look at create_greenland_outline_polygon_DIRTY.r and try using the polygonizer function as opposed to gdal_polygonizeR.
Also, this will only work if you have gdal downloaded locally - if using anaconda python, just type conda install gdal.