[GRASS-user] v.patch drops adjacent polygons

[Markus GRASS]

A. Marcia BARBOSA wrote:
>> What does v.info on the two vectors and the patched vector say in terms
>> of extends and number of objects, in this case only boundaries,
>> centroids, areas, islands?
>>     
>
> $ v.info map=utm10_3132
> [...]
>  |   Number of points:       0               Number of areas:      43280      |
>  |   Number of lines:        0               Number of islands:    1          |
>  |   Number of boundaries:   87166           Number of faces:      0          |
>  |   Number of centroids:    43280           Number of kernels:    0          |
> [...]
>  |         Projection: Lat/Lon                                                |
>  |               N:  72:05:57.201093N    S:  27:05:25.520375N                 |
>  |               E:               12E    W:                 0                 |
>
>
> $ v.info map=utm10_3334
>
>  |   Number of points:       0               Number of areas:      43280      |
>  |   Number of lines:        0               Number of islands:    1          |
>  |   Number of boundaries:   87174           Number of faces:      0          |
>  |   Number of centroids:    43280           Number of kernels:    0          |
>
>  |         Projection: Lat/Lon                                                |
>  |               N:  72:05:57.201093N    S:  27:05:25.520375N                 |
>  |               E:               24E    W:               12E                 |
>
>
> $ v.info map=utm10_31to34     # patched map
>
>  |   Number of points:       0               Number of areas:      85544      |
>  |   Number of lines:        0               Number of islands:    0          |
>  |   Number of boundaries:   174340          Number of faces:      0          |
>  |   Number of centroids:    86560           Number of kernels:    0          |
>
>  |         Projection: Lat/Lon                                                |
>  |               N:  72:05:57.201093N    S:  27:05:25.520375N                 |
>  |               E:               24E    W:                 0                 |
>
>
>   
Doesn't look too bad so far, at least with regard to the extends. If you
zoom to the patched map, something should be displayed.
> So, both input maps have the same number of polygons and centroids (as
> expected), but different number of boundaries - is this normal?
I think no, the number of boundaries in the patched vector should be
identical to the sum of the boundaries in the input vectors.
>  The
> original maps were shapefiles that I imported with v.in.ogr. Should I
> do any additional cleaning of topology? 
Usually not necessary when the shapefiles were cleaned during import by
v.in.ogr which is highly recommended.
> Also he two inputs have an
> island each, but the output has none.
>   
Probably a problem with topology. What is the output of  v.build
map=utm10_31to34?
> The geographic limits are as expected for two longitudinally adjacent
> tiles. And the output has the sum of the number of centroids of the
> two inputs, but less areas...
>   
As with the two islands, probably some incorrect boundaries were the two
tiled original vectors touch. Try v.clean tool=bpol,rmdupl type=boundary
(break polygons, then remove duplicates) on the patched map. Maybe some
snapping is needed. Visual inspection of the area were the two vectors
touch should give you some idea about what's wrong.

Markus M