[GRASS-user] problem with column label using r.cross

GRASS GRASS user
1

Dear GRASS user,

I use r.cross for combining the raster data values of up to 9 layers/maps. The
combinations are stored in column label of each new generated area. But column
label seems to be limited to a specific number of characters, so that some
information get lost in my case of 9 combining maps. My label column only shows
a subpart like:

category 2; category 3; category 2; category 0; category 0; category 9; category

Something is missing in every row. How to solve it?
Thanks a lot for any help!

Regards,
Christian.

----------------------------------------------------------------
This mail was sent through http://webmail.uni-jena.de

2

Christian Schwartze wrote:

I use r.cross for combining the raster data values of up to
9 layers/maps. The combinations are stored in column label of each
new generated area. But column label seems to be limited to a specific
number of characters, so that some information get lost in my case of 9
combining maps. My label column only shows a subpart like:

category 2; category 3; category 2; category 0; category 0; category 9; category

(above ends at the 80th character)

Something is missing in every row. How to solve it?

how are you viewing the cats? where does that cut-short line come from?

I have just done a little test, and it seemed to write the cats file ok.
Both raw if you look in the $MAPSET/cats/$MAPNAME file and printed with r.category. Only issue I found was that the cats file was huge and things like r.info took a long time. As I used fake data I won't worry about that.

my test:

g.region n=4925010 s=4913700 w=589980 e=603000 res=30

for i in `seq 10` ; do
   r.surf.gauss out=surf.gauss_tmp_$i
   r.mapcalc "surf.gauss_$i = int( surf.gauss_tmp_$i )"
   g.remove surf.gauss_tmp_$i --quiet
done

r.cross input=`g.mlist rast pat=surf.gauss_* sep=,` output=surf.gauss.x

?
Hamish

3

Ok and thanks, better to look at first within GRASS if eveything is working fine
:wink: (in cats/ full category values) Because the reduced line comes from QGIS
(selection tool). I forgot to say that following r.cross I do a conversion with
r.to.vect and export the result to a shapefile. So the original label vals were
mapped to dbf-column type "C". But why exporting only 80 chars?

Regards,
Christian.

Zitat von Hamish <hamish_b@yahoo.com>:

Christian Schwartze wrote:
> I use r.cross for combining the raster data values of up to
> 9 layers/maps. The combinations are stored in column label of each
> new generated area. But column label seems to be limited to a specific
> number of characters, so that some information get lost in my case of 9
> combining maps. My label column only shows a subpart like:
>
> category 2; category 3; category 2; category 0; category 0; category 9;
category

(above ends at the 80th character)

> Something is missing in every row. How to solve it?

how are you viewing the cats? where does that cut-short line come from?

I have just done a little test, and it seemed to write the cats file ok.
Both raw if you look in the $MAPSET/cats/$MAPNAME file and printed with
r.category. Only issue I found was that the cats file was huge and things
like r.info took a long time. As I used fake data I won't worry about that.

my test:

g.region n=4925010 s=4913700 w=589980 e=603000 res=30

for i in `seq 10` ; do
   r.surf.gauss out=surf.gauss_tmp_$i
   r.mapcalc "surf.gauss_$i = int( surf.gauss_tmp_$i )"
   g.remove surf.gauss_tmp_$i --quiet
done

r.cross input=`g.mlist rast pat=surf.gauss_* sep=,` output=surf.gauss.x

?
Hamish

----------------------------------------------------------------
This mail was sent through http://webmail.uni-jena.de

4

Just now I read in the gdal description for shapefile driver:
"String fields without an assigned width are treated as 80 characters."
How can I modify that?

Christian.

Zitat von Hamish <hamish_b@yahoo.com>:

Christian Schwartze wrote:
> I use r.cross for combining the raster data values of up to
> 9 layers/maps. The combinations are stored in column label of each
> new generated area. But column label seems to be limited to a specific
> number of characters, so that some information get lost in my case of 9
> combining maps. My label column only shows a subpart like:
>
> category 2; category 3; category 2; category 0; category 0; category 9;
category

(above ends at the 80th character)

> Something is missing in every row. How to solve it?

how are you viewing the cats? where does that cut-short line come from?

I have just done a little test, and it seemed to write the cats file ok.
Both raw if you look in the $MAPSET/cats/$MAPNAME file and printed with
r.category. Only issue I found was that the cats file was huge and things
like r.info took a long time. As I used fake data I won't worry about that.

my test:

g.region n=4925010 s=4913700 w=589980 e=603000 res=30

for i in `seq 10` ; do
   r.surf.gauss out=surf.gauss_tmp_$i
   r.mapcalc "surf.gauss_$i = int( surf.gauss_tmp_$i )"
   g.remove surf.gauss_tmp_$i --quiet
done

r.cross input=`g.mlist rast pat=surf.gauss_* sep=,` output=surf.gauss.x

?
Hamish

----------------------------------------------------------------
This mail was sent through http://webmail.uni-jena.de

5

Christian Schwartze
Ok and thanks, better to look at first within GRASS if
eveything is working fine
:wink: (in cats/ full category values) Because the reduced line
comes from QGIS
(selection tool). I forgot to say that following r.cross I
do a conversion with
r.to.vect and export the result to a shapefile. So the
original label vals were
mapped to dbf-column type "C". But why exporting only 80 chars?

Does v.db.select show the cut string?
which DB backend? (dbf?)

"db.describe -c vect_map" ?

Hamish

6

Zitat von Hamish <hamish_b@yahoo.com>:

> Christian Schwartze
> Ok and thanks, better to look at first within GRASS if
> eveything is working fine
> :wink: (in cats/ full category values) Because the reduced line
> comes from QGIS
> (selection tool). I forgot to say that following r.cross I
> do a conversion with
> r.to.vect and export the result to a shapefile. So the
> original label vals were
> mapped to dbf-column type "C". But why exporting only 80 chars?

Does v.db.select show the cut string?
which DB backend? (dbf?)

v.db.select shows the full length of column label for all data rows - nothing is
missing. I use dbf backend.

"db.describe -c vect_map" ?

ncols:3
Column 1:cat:INTEGER
Column 2:value:INTEGER
Column 3:label:CHARACTER

Christian.

Hamish

----------------------------------------------------------------
This mail was sent through http://webmail.uni-jena.de