What is the source of the coastline database used in the grdlandmask command? Has it changed between GMT 4, 5, and 6?
I was comparing some datasets that I had clipped with the mask from grdlandmask from GMT4 some time ago (using the -Df option) with a map made from Open Street Map in QGIS, and there seems to be a shift of about 400 meters. I was wondering if re-running the calculations with GMT6 would use a different coastline database.
Hi Eric. It is GSHHG version 2.3.7. There have been small changes since GMT 4 (adding missing islands, relocating a few islands, that sort of thing - see GSHHG README for details. So all those minor changes are good. There are major issues in GSHHG in some places, especially NE Greenland. We hope to address that this year but the task is tied into GMT proposals. We would like to move to OpenStreetMap to derive GSHHG 3 but again tied into proposal (non-trivial work that will take someone’s time). The shift you see w.r.t. QGIS is one of the problems in using a 1970s/80s dataset.
Thanks for the explanation. Makes sense that grdlandmask uses the same dataset as pscoast.
The place where I noticed the 400 meter shift was Puerto Rico. I switched to the raster water body dataset that is available with SRTM, as that has the same resolution and grid as the SRTM DEM that I am using for InSAR processing. It is generally consistent with the OpenStreetMap coastline.
Great to hear you are thinking of ways to improve the GSHHG. I can appreciate that it is a non-trivial amount of work.