[GRASS-user] Comparing 2 large Corine Land Cover maps

[Hermann_Peifer]

Hi,

I have 2 Corine Land Cover (CLC) raster maps, from 2000 and 2006. Both maps have the same extent, the same resolution (100m) and use the same 44 CLC categories. The maps are pretty large: 67 000 cols x 58 000 rows = 3 886 000 000 cells.

I'd like to categorise the changes between the 2 layers, according to a decision matrix which I pasted below [1]. Row headers indicate the clc2000 category, column headers represent the clc2006 category. I only pasted the "left half" of the table in order to avoid ugly line breaks.

Question:
Is there a better (and faster) way than doing this comparison by using r.mapcalc with a (nearly) endless if/else chain [2] ?

I also noted that in 99+ % of the cases, both maps are NULL or have an identical CLC category, so I wouldn't need the if/else chain at all. I guess I can use this fact to shortcut the comparison?

Thanks in advance, Hermann
(using GRASS 6.4 on 64-bit Debian Linux)

[1]

## 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 ..
  1 NA 12 31 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  2 11 NA 31 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  3 12 12 NA 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  4 12 12 31 NA 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  5 12 12 31 32 NA 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  6 12 12 31 32 33 NA 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  7 12 12 31 32 33 34 NA 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  8 12 12 31 32 33 34 35 NA 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
  9 12 12 31 32 33 34 35 36 NA 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
10 11 11 31 32 33 34 35 36 37 NA 38 56 56 56 56 56 56 56 56 56 56 56 ..
11 11 11 31 32 33 34 35 36 37 13 NA 56 56 56 56 56 56 56 56 56 56 56 ..
12 21 22 31 32 33 34 35 36 37 13 38 NA 45 43 42 42 42 41 42 41 83 61 ..
13 21 22 31 32 33 34 35 36 37 13 38 43 NA 43 42 42 42 41 42 41 83 61 ..
14 21 22 31 32 33 34 35 36 37 13 38 43 43 NA 42 42 42 41 42 41 83 61 ..
15 21 22 31 32 33 34 35 36 37 13 38 45 45 45 NA 44 44 41 45 45 83 61 ..
16 21 22 31 32 33 34 35 36 37 13 38 45 45 45 44 NA 44 41 45 45 83 61 ..
17 21 22 31 32 33 34 35 36 37 13 38 45 45 45 44 44 NA 41 45 45 83 61 ..
18 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 NA 45 45 83 61 ..
19 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 NA 41 83 61 ..
20 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 45 NA 83 61 ..
21 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 41 52 54 NA 54 ..
22 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 45 45 83 NA ..
23 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 ..
24 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 ..
25 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 ..
26 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
27 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
28 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
29 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
30 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
31 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
32 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
33 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
34 NA NA NA NA NA NA NA NA NA NA 38 NA NA NA NA NA NA NA NA NA NA NA ..
35 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
36 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
37 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
38 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
39 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
40 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
41 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
42 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
43 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
44 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..

[2]

change = \
clc2000 == 1 && clc2006 == 2 ? 12 : \
clc2000 == 1 && clc2006 == 3 ? 31 : \
clc2000 == 1 && clc2006 == 4 ? 32 : \
clc2000 == 1 && clc2006 == 5 ? 33 : \
clc2000 == 1 && clc2006 == 6 ? 34 : \
clc2000 == 1 && clc2006 == 7 ? 35 : \
clc2000 == 1 && clc2006 == 8 ? 36 : \
clc2000 == 1 && clc2006 == 9 ? 37 : \
clc2000 == 1 && clc2006 == 10 ? 13 : \

... (snip away some 1700 if/else rules)

clc2000 == 44 && clc2006 == 30 ? 89 : \
clc2000 == 44 && clc2006 == 31 ? 89 : \
clc2000 == 44 && clc2006 == 32 ? 89 : \
clc2000 == 44 && clc2006 == 35 ? 82 : \
clc2000 == 44 && clc2006 == 36 ? 82 : \
clc2000 == 44 && clc2006 == 37 ? 82 : \
clc2000 == 44 && clc2006 == 38 ? 89 : \
clc2000 == 44 && clc2006 == 39 ? 89 : \
clc2000 == 44 && clc2006 == 42 ? 82 : \
clc2000 == 44 && clc2006 == 43 ? 82 : 0

[ecodiv]

Maybe r.cross can be of help. It creates a cross product of the category variables of a layer, i.e., a map with separate categories for each unique combination of the two CLC maps (the category labels gives you the combination of input maps for each new category).

You can use this new map to create a reclass map with the new values from your decision matrix. To do this, you could for example follow the following steps:

1. Run r.category to get a list with the categories and category labels of the cross product map you just created. You can output this as a text file
2. Use a spreadsheet or something like awk to create a textfile with three columns, one with the new category, and two with the clc2000 and clc2006 categories
3. Find a way to link the CLC combinations in your text file to the new classes given in your decision matrix (use e.g., spreadsheet or database)
4. Use this new table to create the reclassification rules you need in the next step (again, if you know awk this will probably be very easy, but also in R or in a spreadsheet this is easy to do).
5. Create a text file with reclassification rules (see r.reclass how these should look like) and use r.reclass to reclassify the map you just created with r.cross.

I guess there must be easier ways. But in any case, these steps are not very difficult to implement. And in case you have to repeat this type of analysis more often, once you have set up a workflow, it will be easy to repeat (especially if you manage to implement all steps as a script).

On Fri, Jul 22, 2011 at 12:41 PM, Hermann Peifer <peifer@gmx.eu> wrote:

Hi,

I have 2 Corine Land Cover (CLC) raster maps, from 2000 and 2006. Both maps have the same extent, the same resolution (100m) and use the same 44 CLC categories. The maps are pretty large: 67 000 cols x 58 000 rows = 3 886 000 000 cells.

I’d like to categorise the changes between the 2 layers, according to a decision matrix which I pasted below [1]. Row headers indicate the clc2000 category, column headers represent the clc2006 category. I only pasted the “left half” of the table in order to avoid ugly line breaks.

Question:
Is there a better (and faster) way than doing this comparison by using r.mapcalc with a (nearly) endless if/else chain [2] ?

I also noted that in 99+ % of the cases, both maps are NULL or have an identical CLC category, so I wouldn’t need the if/else chain at all. I guess I can use this fact to shortcut the comparison?

Thanks in advance, Hermann
(using GRASS 6.4 on 64-bit Debian Linux)

[1]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 …

1 NA 12 31 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
2 11 NA 31 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
3 12 12 NA 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
4 12 12 31 NA 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
5 12 12 31 32 NA 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
6 12 12 31 32 33 NA 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
7 12 12 31 32 33 34 NA 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
8 12 12 31 32 33 34 35 NA 37 13 38 56 56 56 56 56 56 56 56 56 56 56 …
9 12 12 31 32 33 34 35 36 NA 13 38 56 56 56 56 56 56 56 56 56 56 56 …
10 11 11 31 32 33 34 35 36 37 NA 38 56 56 56 56 56 56 56 56 56 56 56 …
11 11 11 31 32 33 34 35 36 37 13 NA 56 56 56 56 56 56 56 56 56 56 56 …
12 21 22 31 32 33 34 35 36 37 13 38 NA 45 43 42 42 42 41 42 41 83 61 …
13 21 22 31 32 33 34 35 36 37 13 38 43 NA 43 42 42 42 41 42 41 83 61 …
14 21 22 31 32 33 34 35 36 37 13 38 43 43 NA 42 42 42 41 42 41 83 61 …
15 21 22 31 32 33 34 35 36 37 13 38 45 45 45 NA 44 44 41 45 45 83 61 …
16 21 22 31 32 33 34 35 36 37 13 38 45 45 45 44 NA 44 41 45 45 83 61 …
17 21 22 31 32 33 34 35 36 37 13 38 45 45 45 44 44 NA 41 45 45 83 61 …
18 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 NA 45 45 83 61 …
19 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 NA 41 83 61 …
20 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 45 NA 83 61 …
21 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 41 52 54 NA 54 …
22 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 45 45 83 NA …
23 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 …
24 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 …
25 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 …
26 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 …
27 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 …
28 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 …
29 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 …
30 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 …
31 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 …
32 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 …
33 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 …
34 NA NA NA NA NA NA NA NA NA NA 38 NA NA NA NA NA NA NA NA NA NA NA …
35 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
36 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
37 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
38 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
39 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
40 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
41 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
42 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
43 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …
44 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 …

[2]

change =
clc2000 == 1 && clc2006 == 2 ? 12 :
clc2000 == 1 && clc2006 == 3 ? 31 :
clc2000 == 1 && clc2006 == 4 ? 32 :
clc2000 == 1 && clc2006 == 5 ? 33 :
clc2000 == 1 && clc2006 == 6 ? 34 :
clc2000 == 1 && clc2006 == 7 ? 35 :
clc2000 == 1 && clc2006 == 8 ? 36 :
clc2000 == 1 && clc2006 == 9 ? 37 :
clc2000 == 1 && clc2006 == 10 ? 13 : \

… (snip away some 1700 if/else rules)

clc2000 == 44 && clc2006 == 30 ? 89 :
clc2000 == 44 && clc2006 == 31 ? 89 :
clc2000 == 44 && clc2006 == 32 ? 89 :
clc2000 == 44 && clc2006 == 35 ? 82 :
clc2000 == 44 && clc2006 == 36 ? 82 :
clc2000 == 44 && clc2006 == 37 ? 82 :
clc2000 == 44 && clc2006 == 38 ? 89 :
clc2000 == 44 && clc2006 == 39 ? 89 :
clc2000 == 44 && clc2006 == 42 ? 82 :
clc2000 == 44 && clc2006 == 43 ? 82 : 0

---

grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user

[S_Koukoulas]

try
r.cross (Raster>Overlay Rasters>Cross product) for a cross-classification map
and
r.coin (Raster>Reports and Statistics>Mutual category occurrences) for cross tabulation
also r.kappa (Imagery>Reports and statistics>Kappa Analysis) for a comparison

cheers,
s.

[Hermann_Peifer]

Thanks to Paulo and S. for pointing to r.cross. As suggested, I will go for a sequence of r.cross > r.category > awk > r.reclass

r.cross is pretty fast and generates the output map in 16 minutes (r.mapcalc wasn't done after 16 hours..).

Hermann

-------- Original-Nachricht --------

Datum: Sun, 24 Jul 2011 00:15:29 +0200
Von: Paulo van Breugel <p.vanbreugel@gmail.com>
An: Hermann Peifer <peifer@gmx.eu>
CC: grass-user <grass-user@lists.osgeo.org>
Betreff: Re: [GRASS-user] Comparing 2 large Corine Land Cover maps

Maybe r.cross can be of help. It creates a cross product of the category
variables of a layer, i.e., a map with separate categories for each unique
combination of the two CLC maps (the category labels gives you the
combination of input maps for each new category).

You can use this new map to create a reclass map with the new values from
your decision matrix. To do this, you could for example follow the
following
steps:

1) Run r.category to get a list with the categories and category labels of
the cross product map you just created. You can output this as a text file
2) Use a spreadsheet or something like awk to create a textfile with three
columns, one with the new category, and two with the clc2000 and clc2006
categories
3) Find a way to link the CLC combinations in your text file to the new
classes given in your decision matrix (use e.g., spreadsheet or database)
4) Use this new table to create the reclassification rules you need in the
next step (again, if you know awk this will probably be very easy, but
also
in R or in a spreadsheet this is easy to do).
5) Create a text file with reclassification rules (see r.reclass how these
should look like) and use r.reclass to reclassify the map you just created
with r.cross.

I guess there must be easier ways. But in any case, these steps are not
very
difficult to implement. And in case you have to repeat this type of
analysis
more often, once you have set up a workflow, it will be easy to repeat
(especially if you manage to implement all steps as a script).

On Fri, Jul 22, 2011 at 12:41 PM, Hermann Peifer <peifer@gmx.eu> wrote:

> Hi,
>
> I have 2 Corine Land Cover (CLC) raster maps, from 2000 and 2006. Both
maps
> have the same extent, the same resolution (100m) and use the same 44 CLC
> categories. The maps are pretty large: 67 000 cols x 58 000 rows = 3 886
000
> 000 cells.
>
> I'd like to categorise the changes between the 2 layers, according to a
> decision matrix which I pasted below [1]. Row headers indicate the
clc2000
> category, column headers represent the clc2006 category. I only pasted
the
> "left half" of the table in order to avoid ugly line breaks.
>
>
> Question:
> Is there a better (and faster) way than doing this comparison by using
> r.mapcalc with a (nearly) endless if/else chain [2] ?
>
> I also noted that in 99+ % of the cases, both maps are NULL or have an
> identical CLC category, so I wouldn't need the if/else chain at all. I
guess
> I can use this fact to shortcut the comparison?
>
>
> Thanks in advance, Hermann
> (using GRASS 6.4 on 64-bit Debian Linux)
>
> [1]
>
> ## 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 ..
> 1 NA 12 31 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 2 11 NA 31 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 3 12 12 NA 32 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 4 12 12 31 NA 33 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 5 12 12 31 32 NA 34 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 6 12 12 31 32 33 NA 35 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 7 12 12 31 32 33 34 NA 36 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 8 12 12 31 32 33 34 35 NA 37 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 9 12 12 31 32 33 34 35 36 NA 13 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 10 11 11 31 32 33 34 35 36 37 NA 38 56 56 56 56 56 56 56 56 56 56 56 ..
> 11 11 11 31 32 33 34 35 36 37 13 NA 56 56 56 56 56 56 56 56 56 56 56 ..
> 12 21 22 31 32 33 34 35 36 37 13 38 NA 45 43 42 42 42 41 42 41 83 61 ..
> 13 21 22 31 32 33 34 35 36 37 13 38 43 NA 43 42 42 42 41 42 41 83 61 ..
> 14 21 22 31 32 33 34 35 36 37 13 38 43 43 NA 42 42 42 41 42 41 83 61 ..
> 15 21 22 31 32 33 34 35 36 37 13 38 45 45 45 NA 44 44 41 45 45 83 61 ..
> 16 21 22 31 32 33 34 35 36 37 13 38 45 45 45 44 NA 44 41 45 45 83 61 ..
> 17 21 22 31 32 33 34 35 36 37 13 38 45 45 45 44 44 NA 41 45 45 83 61 ..
> 18 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 NA 45 45 83 61 ..
> 19 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 NA 41 83 61 ..
> 20 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 45 NA 83 61 ..
> 21 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 41 52 54 NA 54 ..
> 22 21 22 31 32 33 34 35 36 37 13 38 45 45 45 42 42 42 41 45 45 83 NA ..
> 23 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 ..
> 24 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 ..
> 25 21 22 31 32 33 34 35 36 37 13 38 51 51 51 51 51 51 53 51 53 53 53 ..
> 26 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
> 27 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
> 28 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
> 29 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 54 ..
> 30 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
> 31 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
> 32 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
> 33 21 22 31 32 33 34 35 36 37 13 38 52 52 52 52 52 52 54 52 54 54 53 ..
> 34 NA NA NA NA NA NA NA NA NA NA 38 NA NA NA NA NA NA NA NA NA NA NA ..
> 35 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 36 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 37 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 38 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 39 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 40 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 41 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 42 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 43 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
> 44 21 22 31 32 33 34 35 36 37 13 38 55 55 55 55 55 55 55 55 55 55 55 ..
>
>
> [2]
>
> change = \
> clc2000 == 1 && clc2006 == 2 ? 12 : \
> clc2000 == 1 && clc2006 == 3 ? 31 : \
> clc2000 == 1 && clc2006 == 4 ? 32 : \
> clc2000 == 1 && clc2006 == 5 ? 33 : \
> clc2000 == 1 && clc2006 == 6 ? 34 : \
> clc2000 == 1 && clc2006 == 7 ? 35 : \
> clc2000 == 1 && clc2006 == 8 ? 36 : \
> clc2000 == 1 && clc2006 == 9 ? 37 : \
> clc2000 == 1 && clc2006 == 10 ? 13 : \
>
> ... (snip away some 1700 if/else rules)
>
> clc2000 == 44 && clc2006 == 30 ? 89 : \
> clc2000 == 44 && clc2006 == 31 ? 89 : \
> clc2000 == 44 && clc2006 == 32 ? 89 : \
> clc2000 == 44 && clc2006 == 35 ? 82 : \
> clc2000 == 44 && clc2006 == 36 ? 82 : \
> clc2000 == 44 && clc2006 == 37 ? 82 : \
> clc2000 == 44 && clc2006 == 38 ? 89 : \
> clc2000 == 44 && clc2006 == 39 ? 89 : \
> clc2000 == 44 && clc2006 == 42 ? 82 : \
> clc2000 == 44 && clc2006 == 43 ? 82 : 0
> ______________________________**_________________
> grass-user mailing list
> grass-user@lists.osgeo.org
>
http://lists.osgeo.org/**mailman/listinfo/grass-user&lt;http://lists.osgeo.org/mailman/listinfo/grass-user&gt;
>

[Glynn_Clements1]

Hermann Peifer wrote:

I have 2 Corine Land Cover (CLC) raster maps, from 2000 and 2006. Both
maps have the same extent, the same resolution (100m) and use the same
44 CLC categories. The maps are pretty large: 67 000 cols x 58 000 rows
= 3 886 000 000 cells.

I'd like to categorise the changes between the 2 layers, according to a
decision matrix which I pasted below [1]. Row headers indicate the
clc2000 category, column headers represent the clc2006 category. I only
pasted the "left half" of the table in order to avoid ugly line breaks.

Question:
Is there a better (and faster) way than doing this comparison by using
r.mapcalc with a (nearly) endless if/else chain [2] ?

As Paulo suggests, r.cross and r.reclass. Alternatively, the r.cross
step could be done with r.mapcalc as e.g.:

  r.mapcalc 'result = map1 * 100 + map2'

I also noted that in 99+ % of the cases, both maps are NULL or have an
identical CLC category, so I wouldn't need the if/else chain at all. I
guess I can use this fact to shortcut the comparison?

r.mapcalc doesn't do short-circuit evaluation, which is why long
"if-else" chains are slow.

--
Glynn Clements <glynn@gclements.plus.com>

[Hermann_Peifer]

Glynn Clements wrote:

Alternatively, the r.cross
step could be done with r.mapcalc as e.g.:

  r.mapcalc 'result = map1 * 100 + map2'

Thanks for the hint. I should have thought about it myself.

r.mapcalc doesn't do short-circuit evaluation, which is why long
"if-else" chains are slow.

Which is a pity in my case. But luckily, I just learned how to avoid long if-else chains.

Thanks for your time,

Hermann

[Hermann_Peifer]

On 24/07/2011 00:15, Paulo van Breugel wrote:

Maybe r.cross can be of help. It creates a cross product of the category
variables of a layer, i.e., a map with separate categories for each
unique combination of the two CLC maps (the category labels gives you
the combination of input maps for each new category).

You can use this new map to create a reclass map with the new values
from your decision matrix. To do this, you could for example follow the
following steps:

1) Run r.category to get a list with the categories and category labels
of the cross product map you just created. You can output this as a text
file

Sorry for coming back to this older mail. My question is: rather than running the somewhat time-consuming r.category to get a list with the categories and their labels, can I not simply use:

cat cats/crossmap # ..and there I have my text file with the labels?

(Given that I want all categories for the full map extent in the native resolution.)

Hermann