GMT grid maps

GMT grid maps are basically netcdf files. Gstat only includes GMT support when it is linked with the netcdf library. This library is detected automatically by the configure script (section 5.6), or it is added when configure is invoked as

./configure -with-netcdf

GMT map support was contributed by Konstantin Malakhanov who wrote about it:

``And last but not least: I write here all limitations of using GMT grids with GSTAT :-

1.

GMT grids can be centered at pixels or at nodes. GSTAT grids are centered at pixels, so node-wise GSTAT grids will be converted to pixel-centered. GMT grids from GSTAT are always pixel-centered. Convertation could be made in two ways: either decrease the number of rows and columns by one and set pixel values to mean (or median, or what you like at most) value of 4 nodes (this changes values, but preserves boundaries of grid) or extent grid limits to half cell size to west/east/north/south and use nodes as centers of new pixels (this preserves values, but slightly changes the limits of grids). I implemented the second way, so if you have GMT node-centered grid as mask, then the extensions of result grid from gstat will be one cell size bigger in X- and Y-directions!

2.

GMT grids can have multiplication factor and an add offset for z-values. As GSTAT grid definition does not allow for that, GMT grid values with factor different from 1.0 and/or value offset different from 0.0 will be accordingly transformed during the loading. (Comment: for reasons I cannot understand GMT grids have sometimes factor 0.0 which makes no sense. So factor==0.0 will be treated as 1.0). GSTAT grids in GMT format always have factor== 1.0 and value offset==0.0.

3.

GMT system allows for rectangular coordinate system or for geographical projections, but there is no way to detect it from grid itself (in GMT commands , projection is almost always supplied as one of arguments by user). So the way GMT grids are treated is defined by user and not stored in a grid. That means that using GSTAT for grids which are supposed to have longitude/lattitude coordinates WILL give results, but these results are useless as spheroid of Earth is not taken into account and it means by no way that GSTAT can interpolate/simulate over sphere in geographical projections (if one needs such things, take a look at Spherekit at http://www.ncgia.ucsb.edu/pubs/spherekit). So I didn't follow this branch further.

4.

GMT grid definition has fields for names of x-,y-,z-units. These fields are ignored at reading and will be set to " " in the result grid.

5.

GMT grids can have complex z-values. This is neither checked for nor used!''