[GPlates-discuss] Exporting high-resolution reconstructed rasters

[John Cannon]

Please note, this thread is continued at https://protect-au.mimecast.com/s/RMxeC0YKPviMkRr1CwolGl?domain=discourse.gplates.org


From: GPlates-discuss <gplates-discuss-bounces at mailman.sydney.edu.au> On Behalf Of John Cannon
Sent: Thursday, 20 February 2020 8:15 PM
To: Thomas Schaap <thomas.schaap at utas.edu.au>; GPlates general discussion mailing list <gplates-discuss at mailman.sydney.edu.au>
Subject: Re: [GPlates-discuss] Exporting high-resolution reconstructed rasters

Might be graphics driver related. Can you send me a repro privately (the original raster, export bounds, etc; and your OS version and graphics card model)?  I can look into it next week when back from hols, just to be sure it's not a GPlates bug.

Sent while away from keyboard

________________________________
From: Thomas Schaap <thomas.schaap at utas.edu.au<mailto:thomas.schaap at utas.edu.au>>
Sent: Wednesday, 19 February 2020, 10:55 am
To: John Cannon; GPlates general discussion mailing list
Subject: Re: Exporting high-resolution reconstructed rasters


Hi again,

Thank you for your inputs. As always when I ask for help on these problems I found that I hadn't been doing it properly. I wasn't properly setting the exported region (it's not a global dataset). This significantly cut down the size of the export to the point where GPlates will now allow me to export the data at its original resolution.

The trouble I'm having now is that the exported raster looks to contain only certain parts of the data, as if the export process has cut the raster into segments and then left some of them out. It's hard to describe so I've attached a screenshot to describe what I'm talking about. The grayscale area is the reconstructed raster I'm trying to export as displayed in GPlates, and the green areas are the exported raster when I import it back in (it's the same result when I look at it in other GIS packages). This doesn't happen for lower resolution exports.

Has anyone seen this behaviour before? Any advice?

Thanks,
Tom


Thomas Schaap

PhD Candidate | Casual Academic Staff

CODES | Earth Sciences | Rm. 460

School of Natural Sciences

University of Tasmania
[cid:image001.gif at 01D609FD.6C2A0F80]

________________________________
From: GPlates-discuss <gplates-discuss-bounces at mailman.sydney.edu.au<mailto:gplates-discuss-bounces at mailman.sydney.edu.au>> on behalf of Michael Chin <michael.chin at sydney.edu.au<mailto:michael.chin at sydney.edu.au>>
Sent: 19 February 2020 13:24
To: John Cannon <john.cannon at sydney.edu.au<mailto:john.cannon at sydney.edu.au>>; GPlates general discussion mailing list <gplates-discuss at mailman.sydney.edu.au<mailto:gplates-discuss at mailman.sydney.edu.au>>
Subject: Re: [GPlates-discuss] Exporting high-resolution reconstructed rasters

Hi Tom,

Could you please double check the resolution of your original raster? If the tif file is only 43M, it is unlikely the resolution can be so high as 0.0005 degree, which is a 720000x360000 raster, unless the raster contains very sparse data. If the resolution of the original raster is not that high, it is not necessary to export it as 0.0005 degree. The export can only be as good as the original one.
________________________________
From: GPlates-discuss <gplates-discuss-bounces at mailman.sydney.edu.au<mailto:gplates-discuss-bounces at mailman.sydney.edu.au>> on behalf of John Cannon <john.cannon at sydney.edu.au<mailto:john.cannon at sydney.edu.au>>
Sent: Wednesday, 19 February 2020 11:53 AM
To: GPlates general discussion mailing list <gplates-discuss at mailman.sydney.edu.au<mailto:gplates-discuss at mailman.sydney.edu.au>>
Subject: Re: [GPlates-discuss] Exporting high-resolution reconstructed rasters

Hi Tom,

Wow, that's a big raster!  I'm surprised GPlates didn't trip up reconstructing that.

When exporting rasters we currently pass raster data to the image drivers as a single 'uncompressed' memory buffer. Hence the terabyte. Whereas the original tiff is highly compressed. Maybe with some of the 'tiled' image formats we could export differently in GPlates in the future, but that would involve some rearranging of things and would be quite low priority.

The 0.01 degree you successfully exported is near the 32-bit limit. You might be able to go higher with a 64-bit Windows build (assuming you're using a 32-bit build; Mac and Linux are already 64-bit).

Regards,
John

Sent while away from keyboard

________________________________
From: GPlates-discuss <gplates-discuss-bounces at mailman.sydney.edu.au<mailto:gplates-discuss-bounces at mailman.sydney.edu.au>> on behalf of Thomas Schaap <thomas.schaap at utas.edu.au<mailto:thomas.schaap at utas.edu.au>>
Sent: Tuesday, 18 February 2020, 9:08 am
To: GPlates general discussion mailing list
Subject: [GPlates-discuss] Exporting high-resolution reconstructed rasters


Hi everyone,

I am trying to export a high-resolution (0.0005 degree cellsize) numerical raster which I have reconstructed using GPlates. I would ideally like to keep this resolution, but the GPlates console gives me this:

ERROR 2: Cannot allocate 1036800000000 bytes on this platform.
[Warning] Unable to create in-memory dataset for writing rasters.

Now please correct me if I misunderstand what's going on here, but the original TIF file was only 43 megabytes, yet GPlates is asking for 1 terabyte of memory? The lowest cellsize I have been allowed to export is 0.01 degrees.

Is there something I am misunderstanding about the raster export process? Is there some way around this?

Thanks in advance!

Tom


Thomas Schaap

PhD Candidate | Casual Academic Staff

CODES | Earth Sciences | Rm. 460

School of Natural Sciences

University of Tasmania
[cid:image001.gif at 01D609FD.6C2A0F80]


University of Tasmania Electronic Communications Policy (December, 2014).
This email is confidential, and is for the intended recipient only. Access, disclosure, copying, distribution, or reliance on any of it by anyone outside the intended recipient organisation is prohibited and may be a criminal offence. Please delete if obtained in error and email confirmation to the sender. The views expressed in this email are not necessarily the views of the University of Tasmania, unless clearly intended otherwise.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.sydney.edu.au/pipermail/gplates-discuss/attachments/20200403/ce78ff20/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.gif
Type: image/gif
Size: 7449 bytes
Desc: image001.gif
URL: <http://mailman.sydney.edu.au/pipermail/gplates-discuss/attachments/20200403/ce78ff20/attachment-0001.gif>