<div dir="ltr">Bonjour,<div><br></div><div>For testing, I would suggest that you operate on a decimated (sub-sampled) array.</div><div>This will make for quick turn around and verify that the options you are using are correct</div><div><br></div><div> f= addfile("...","r")</div><div><br></div><div> jy = 10</div><div> ix = 10</div><div> x= f->X(::jy,::ix) ; subsampled array</div><div> lat = f->LAT(::jy,::ix)</div><div> lon = .....</div><div>etc</div><div><br></div><div>---</div><div>Currently, NCL's implementation of the ESMF regrid software uses a single processor. The underlying code is capable of parallel operation but (to my knowledge) this has not been implemented.</div><div><br></div><div>---</div><div>Also, when sending to ncl-talk, it is nice to have a 'printVarSummary(...)' of the variable(s) you are using.</div><div>Or a ncl_filedump ...</div><div><br></div><div>Bon Chance!</div><div><br></div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Thu, Jul 16, 2015 at 8:44 AM, Gabriele Arduini <span dir="ltr"><<a href="mailto:Gabriele.Arduini@legi.grenoble-inp.fr" target="_blank">Gabriele.Arduini@legi.grenoble-inp.fr</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Dear ncl-support,<br>
<br>
I am trying to regrid a large high-resolution dataset (7200 lat x 7200<br>
lon) on a coarser resolution grid using<br>
the ESMF_regrid tool.<br>
Everything is going well, but during the weight generation (that takes a<br>
huge amount of time...)<br>
I receive this error::<br>
<br>
(0) --------------------------------------------------<br>
(0) ESMF_regrid_gen_weights: the following command is about to be<br>
executed on the system:<br>
(0) 'ESMF_RegridWeightGen --source <a href="http://source_grid.nc" rel="noreferrer" target="_blank">source_grid.nc</a> --destination<br>
<a href="http://dest_grid.nc" rel="noreferrer" target="_blank">dest_grid.nc</a> --weight weight_file --method patch --src_regional<br>
--dst_regional -i'<br>
(0) --------------------------------------------------<br>
(0) ESMF_regrid_gen_weights: output from 'ESMF_RegridWeightGen':<br>
(0) Starting weight generation with these inputs:<br>
(1) Source File: <a href="http://source_grid.nc" rel="noreferrer" target="_blank">source_grid.nc</a><br>
(2) Destination File: <a href="http://dest_grid.nc" rel="noreferrer" target="_blank">dest_grid.nc</a><br>
(3) Weight File: weight_file<br>
(4) Source File is in SCRIP format<br>
(5) Source Grid is a regional grid<br>
(6) Source Grid is a logically rectangular grid<br>
(7) Destination File is in SCRIP format<br>
(8) Destination Grid is a regional grid<br>
(9) Destination Grid is a logically rectangular grid<br>
(10) Regrid Method: patch<br>
(11) Pole option: NONE<br>
(12) Ignore unmapped destination points<br>
(13) Norm Type: dstarea<br>
(14)<br>
(15) NetCDF Error: weight_file : NetCDF: One or more variable sizes<br>
violate format constraints<br>
(16) ERROR: Problem on processor 0 . Please see the<br>
PET*.RegridWeightGen.Log files for a traceback.<br>
(0) --------------------------------------------------<br>
(0) ESMF_regrid_gen_weights: 'ESMF_RegridWeightGen' was not successful.<br>
<br>
<br>
I understand the error ( my dataset is too big being written in NetCDF-3<br>
file --> 7200 * 7200 * 8 > 4Gb), but I don't know how I can resolve it.<br>
Looking on the web, I saw that you had similar problems in the past.<br>
Does a solution exist?<br>
<br>
Thank you in advance for your precious help.<br>
Regards,<br>
Gabriele Arduini<br>
_______________________________________________<br>
ncl-talk mailing list<br>
<a href="mailto:ncl-talk@ucar.edu">ncl-talk@ucar.edu</a><br>
List instructions, subscriber options, unsubscribe:<br>
<a href="http://mailman.ucar.edu/mailman/listinfo/ncl-talk" rel="noreferrer" target="_blank">http://mailman.ucar.edu/mailman/listinfo/ncl-talk</a><br>
</blockquote></div><br></div>