Re: GMT 4: combining GRD files into one FILE with several LAYERS

"Walter (HF) Smith" <[email protected]>
Newsgroups gmane.comp.gis.gmt.user
Message-ID <[email protected]>
JJ,

Although the number crunching might be parallelizable, it sounds like the I/O of the data to/from tiles is going to be the bottleneck. If you are going to the trouble of building a data structure to hold everything, you might consider also organizing the data by tiles that group lat,lon coordinates, such as by using a quadtree structure, rather than using a flat array of lon,lat as in a conventional GMT grid file. Seems to me that conventional storage would separate by large strides the elements you need to pull together for computation.

Walter

On Feb 11, 2014, at 6:04 PM, Joaquim Luis <[email protected]> wrote:

> I never tried it but it's not impossible that you can do what you want by forcing the writing via GDAL.
> Note that an alternative to a stack of 2D grids would be to have 256 grids written as individual grids inside the netCDF file (SUBDATASETS in GDAL parlance).
> GMT could easily read those individual arrays and perhaps you could force the updating via a GDAL write. See this for an example of syntax for reading
> 
> http://gmtrac.soest.hawaii.edu/doc/5.1.0/GMT_Docs.html#reading-more-complex-multi-band-images-or-grids
> 
>> This data is essentially something along the lines of
>> 
>> 	layer 1 = topogrpahy
>> 	layer 2 = gravity
>> 	layer 43= FFT(gravity) + sqrt (topo)  (or something crazy like that)...
>> 
>> and so on.
>> 
>> In effect a global gird of all known “measurements", not so much “n dimensional” although that is certainly a good way to think of it.
>> 
>> Does that change any facts, your opinion, etc…
>> 
>> -jj
>> 
>> 
>> On Feb 10, 2014, at 5:39 PM, Paul Wessel<[email protected]>  wrote:
>> 
>>> Hi JJ_
>>> 
>>> GMT can read from layer "a" but does not write 3-D grids.  Given that I would not try to do a 256-layer file.
>>> 
>>> -p
>>> 
>>> On Feb 10, 2014, at 9:46 AM, Becker, Joseph J.<[email protected]>  wrote:
>>> 
>>>> I have 256 global grids at 15 second resolution; e.g. ~16 GB grd files * 256 grids =  4 TB of data.
>>>> 
>>>> I want to automate some processing that uses tiny patches (16x16 nodes) of each and every one of these 256 files. And I step that over the globe.
>>>> 
>>>> Off hand it seems like it might make sense (code re-use and simplicity) to combine all that data into a single grid with 256 layers.
>>>> 
>>>> Is this practical?
>>>> 
>>>> Anyone have some examples of doing things like this:
>>>> 
>>>> 	grdmath on layer "a" writing result to layer "b"
>>>> 
>>>> 	grdmask xy files>  "layer"
>>>> 
>>>> and so on.
>>>> 
>>>> thanks!
>>>> 
>>>> 
>>>> -jj
>>>> 
>>>> To unsubscribe, send the message "signoff gmt-help" to [email protected]
>>> To unsubscribe, send the message "signoff gmt-help" to [email protected]
>> To unsubscribe, send the message "signoff gmt-help" to [email protected]
>> 
>> 
> 
> To unsubscribe, send the message "signoff gmt-help" to [email protected]

To unsubscribe, send the message "signoff gmt-help" to [email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.