[GRASS-user] v.in.ascii error: G_realloc

[Praktikant/in2 ARP]

>-----Ursprüngliche Nachricht-----
>Von: Maciej Sieczka [mailto:tutey at o2.pl]
>Gesendet am: Montag, 7. August 2006 13:21
>An: Praktikant/in2 ARP
>Cc: 'grassuser at grass.itc.it'
>Betreff: Re: [GRASS-user] v.in.ascii error: G_realloc
>
>Praktikant/in2 ARP napisa?(a):
>
>> I have about 50 tiles with geodetic points. These geodetic points should
be
>> transformed into a Digital terrain model (DTM) by v.surf.rst.
>
>Can you patch those 50 tiles points outside Grass?
>
>If they are only plain text files, formatted like
>
>x1 y1 z1
>x2 y2 z2
>x3 y3 z3
>...
>
>it will be very easy on Unix:
>
>cat patch49 >> patch50
>cat patch48 >> patch50
>cat patch47 >> patch50
>...
>
>and so forth till they all are in a one file, the "patch50".
>
>Now import that huge patch50 as 3D, not building topology (v.in.ascii -zb).
>
>Set the region *manually* to cover the whole vector patch50 ('g.region
>vect=' will not work with topologyless vector as you know, so you must
>use 'g.region n= s= e= w= res=' explicitely).
>
>Interpoloate dem out of your 3D points (v.surf.rst layer=0).

Thank you for the help. I've done everything like Maciek suggested. So
everything is working fine until using v.surf.rst.
I have one big vector file in Grass with about 11Million points. But when I
try to interpolate the file
with v.surf.rst the same error occurs: 

ERROR: G_realloc: out of memory 

Seems that the file is too big for v.surf.rst. 7952735 points are used for
the interpolation an 60521936bytes are used for the temporary files.
v.surf.idw as alternative doesn`t work because of the missing topology.
It is not possible to make the file smaller.

So any suggestions will be very welcome :-)

Best regards 

Michael
>
>> If somebody wants a DTM for a defined extract of the area the DTM should
be calculated
>> exactly for this area. Everything should be automated by a PHP-Script. I
>> thought about first identifing the needed tiles then patch these tiles
with
>> v.patch, setting the region to the defined extract (g.region) and as last
>> point transform the points by v.surf.rst into a DTM.
>
>It will be not optimal at all to interpolate new dem each time a subset
>is required. Better interploate full dem only once (as I described
>above) and only adjust region settings as needed, for each query your
>PHP script gets, with g.region, then use 'r.mapcalc current_subset=dem'
>to extract that particular subset from the whole dem. Easier and much
>faster.
>
>Best,
>Maciek