GRASS and GMT

GMT (The Generic Mapping Tools)

GMT (Generic Mapping Tools) is a Free software package for creating publication quality cartography.

GMT homepage: http://gmt.soest.hawaii.edu

Importing from GMT

Raster

• Use the grdinfo utility from GMT to check the Grid file format. See the grdreformat man page for an explanation of the codes.
  • Codes starting with n are "new format" (NetCDF) which can be read by GDAL tools and imported with the r.in.gdal module.
  • Codes starting with b are "old format" which can be imported with the r.in.bin module.

Old style binary grd

• Import using r.in.bin with the -h flag.

Note: (in case you missed it) r.in.bin -h was written to be used with the old style GMT grd format.

• Grid file format codes (see above) ending in f contain floating point data and require that r.in.bin be passed the -f flag. Format codes ending in d contain double precision floating point data and require the -d flag.

New style NetCDF grd

• r.in.gdal is used to import the new NetCDF GMT grd format.

See the GDAL netCDF driver page.

• If for some reason r.in.gdal doesn't work (e.g. your copy of GDAL wasn't build with GMT/NetCDF support) you can convert the new-style grid file into the old-style grid format and then proceed to import with the r.in.bin -h method as given above. Use the GMT grdreformat tool to perform the conversion:

grdreformat in.grd out.grd=bf

Sometimes the new-style NetCDF grid contains multiple layers of data and you'll have to explicitly specify the layer in the input grid that you want. For example, if the layer is called "z":

grdreformat in.grd?z out.grd=bf

• Examination of the grdinfo utility's output may give you some help, as will the output of gdalinfo.

For an example of importing a GMT grid into GRASS, see this page.

Vectors

• Newer versions of OGR support GMT vectors for read/write. Import with the v.in.ogr module.

Coloring

Importing CPT color tables

• Import GMT style .cpt color table files with the r.cpt2grass script on the Add-ons page.
  • r.cpt2grass - Import GMT color rules

For a large collection of GMT .cpt files see http://sview01.wiredworkplace.net/pub/cpt-city/

Exporting GRASS maps to GMT

• Overview: http://169.237.35.250/~dylan/grass_user_group/#GMT_and_GRASS-overview

(Supplied by the GRASS Users Group of Davis, California)

Currently there are several *.out.GMT permutations on the Add-ons page, in several different languages (bash, python, etc.), and each of which with relative pros/cons. An effort to unify these approaches would save much of the current difficulties in moving complex raster+vector data into a GMT-friendly format. A simple road map toward this goal is outlined below.

• v.out.gmt - Vector export for GMT (rewrite to use new OGR r/w support?)
• r.out.gmt - Raster export for GMT
• r.out.gmt2 - Alternative raster export for GMT

To aid custom scripting, g.region in GRASS 6.5+ supports the -t flag to output the current region in GMT's W/E/S/N region format.

Vector

• Conversion of GRASS vector data to GMT compatible ascii files

There is currently an effort (with some funding!), see some of the chatter on the GRASS and GMT mailing lists: GRASS-list GMT-help

• OGR now supports GMT vector (multiline) output, so it is easy to convert data to GMT format for plotting. Use the v.out.ogr module.

• If outputting areas be sure to explicitly request them, otherwise data attributes may not be transfered. The -c flag limits the export to areas with a centroid (i.e. so holes don't get filled)

For example:

v.out.ogr -c input=RBD_F1v3 dsn=RBD_F1v3.gmt format=GMT type=area

Examples

Export and plot areas

Export area vectors from a non-geographic (projected) location. Uses the "trn_sites" map from the Spearfish sample dataset.

MAP=trn_sites

v.out.ogr -c in=$MAP dsn=$MAP type=area format=GMT

# get map bounds
eval `v.info -g $MAP`
# ugly hack to round outward to the next whole kilometer
w=`expr \( \( $(echo $west | cut -f1 -d.) - 1000 \) / 1000 \) \* 1000`
e=`expr \( \( $(echo $east | cut -f1 -d.) + 1000 \) / 1000 \) \* 1000`
s=`expr \( \( $(echo $south | cut -f1 -d.) - 1000 \) / 1000 \) \* 1000`
n=`expr \( \( $(echo $north | cut -f1 -d.) + 1000 \) / 1000 \) \* 1000`

# GMT plotting command
psxy $MAP.gmt \
   -R$w/$e/$s/$n \
   -JXh \
   -G220/220/220 -W -M \
   -B2000/2000:."Map of $MAP":WeSn \
   --D_FORMAT='%.15g' > $MAP.ps

psxy options explained:

 -R$w/$e/$s/$n  # bounding box
 -JXh           # map projection is non-geographic, set map-box width to fill a landscape page
 -G220/220/220  # area fill color to a light grey (R/G/B, 0-255) 
 -W             # draw area boundary
 -M             # break polygons
 -B2000:."Map of $MAP":WeSn   # tick interval : title : tick numbers on sides with capital letters
 --D_FORMAT='%.15g'  # grid reference tick number formatting

If exporting lat/lon data you might skip the rounding outwards hack shown above and change the psxy options as follows:

  -Jm15
  -B30m/15m:."Map Title":WeSn
  --PLOT_DEGREE_FORMAT=ddd:mmA

See GMT's psxy, psbasemap, and gmtdefaults man pages for further details.

Overlay lines

Export a vector containing lines:

v.out.ogr in=roads dsn=roads format=GMT type=line

GMT commands are same as above, but add the -K flag the first time you run psxy to keep the PostScript file open, then the second time you run psxy you can simplify the options a bit and give it the -O flag to tell it that this layer will be an overlay, not a new map. Then append (>> instead of >) to the end of the existing PostScript file.

psxy $MAP.gmt \
   -R$w/$e/$s/$n \
   -JXh \
   -G220/240/220 -W -M \
   -B2000/2000:."Map of $MAP":WeSn \
   --D_FORMAT='%.15g' \
   -K > $MAP.ps

psxy roads.gmt -R -J -M -O >> ${MAP}.ps

Raster

• Conversion of GRASS raster data to GMT compatible binary grids

A combination of r.out.bin | xyz2grd can accomplish this. Several attempts at generalizing this procedure have been proposed: r.out.gmt.py, r.out.gmt (Hamish and Dylan), r.out.gmt.sh (Dylan, based Hamish's work).

• Imagery data e.g. aerial photos

For multiband imagery data, the following commands will export the different colour bands:

r.mapcalc "image.red=r#image; image.green=g#image; image.blue=b#image"
r.out.bin -h input=image.red output=image.red.grd
r.out.bin -h input=image.green output=image.green.grd
r.out.bin -h input=image.blue output=image.blue.grd

They can then be plotted with:

grdimage image.red.grd image.green.grd image.blue.grd -J -R -B ...etc.

Examples

The r.out.gmt script is only set up to export CELL maps at present. See the script comments for details on how to export a FCELL or DCELL map.

Export a raster and overlay lines

This example uses the Spearfish dataset (UTM projection).

MAP=elevation.dem
g.region rast=$MAP

r.out.gmt $MAP -p

# set up plot (from instructions in last command)
psbasemap -JX7i/5.153239i -R590010/609000/4914020/4928000 \
   -Bpf300a3000:"Easting (m)":/f300a3000:"Northing (m)"::."elevation.dem":WeSn \
   -Xc -Yc -U"scale 1:106802" \
   --D_FORMAT='%.15g' \
   -K -V > $MAP.ps

# overplot exported raster grid
grdimage $MAP.grd=2 -Celevation.dem.cpt -JX7i/5.153239i \
   -R -K -O -V >> $MAP.ps


# export roads vector
v.out.ogr in=roads dsn=roads format=GMT type=line

# overplot roads
psxy roads.gmt -R -J -M -O >> $MAP.ps

Coloring

• Conversion of GRASS raster color data into GMT compatible CPT files

David Finlayson's r.out.gmt.py does a nice job of this. Once we decide on an optimal language to implement the routines in this may need translation.

Symbology

• Automatic conversion of symbology data stored in a gis.m or QGIS saved state to GMT options

Ideas expressed on various mailing list, haven't seem much since. It should be a relatively simple excercise in XML parsing to convert symbology stored in a QGIS project file into something that GMT can use.

Development plans

• General approach:

Since GMT relies on a sequence of specialized programs to "build-up" a postscript file, some thought must be put into how the conversion should take place. As usual, form should follow function- maximum flexibility, robustness, and accuracy being primary objectives. However, a simple means of creating high quality 2D maps would be a tremendous (I think) addition to the GRASS toolset. Especially since this is something frequently cited by critics. --DylanBeaudette 02:47, 10 December 2006 (CET)

1. should we continue down the well troden path of single-use, highly efficient programs for the various conversion steps: i.e v.out.GMT, r.out.GMT, etc.?

2. should there be a unified approach to the process: something akin to ps.map - GMT.map ?

Tutorials

• Dane DEM uzyskane z radarowej misji promu kosmicznego - SRTM-3: Tutorial in Polish