THORSTEN
> We are using MapInfo Prof. 4.5.2 (German version). For a digital soil map,
> which we had purchased in the Arc/Info e00 format and had translated to a
> MapInfo relation, the utility MapCheck revealed about 20 topology errors
> (intersections, bow ties, figure 8) which we corrected manually.
> So finally MapCheck did not reveal any further faults.
> However, when splitting this relation by any other relation (MapCheck
reveals
> no faults for the relations we are using to split) we get the error
message
>
> "Fehler beim verkn�pfen der Objekte" (in German), which may sound
> like "Errors occured whilst splitting/combining objects" in English.
>
> As you can imagine, we urgently need the splitted relation. So
> can anybody of you help to overcome this problems?
You have there a perfect example of the situation pre-version 5.5. A "good"
looking object collection that, when submitted to graphic manipulations,
will yield "EOO" (errors overlaying objects). There are no ways you can
systematically avoid them with versions <5.5.
I have written a small app (polynode) that detect potentially troublesome
areas, but if you do the suggested corrections (by hand), there is no
guarantee that the set of regions obtained from one manipulation (let say
splitting) will be trouble free for another one. You will have to go through
the process again.
A solution to your splitting problem would be to find someone with 5.5 to
run that operation for you (even a beta version would do, but hurry up
because it had a time stamp, mine was getting defunct on July 1)
> In addition, we possess a landuse relation for which MapCheck reveals more
than
> 120 topology errors. Is there any way to correct those automatically? If
> necessary we will not hesitate to purchase useful software.
You may wish to leave a program decide for you the corrections to bring to
trouble areas. It is an understandable attitude when one does not care
knowing what are the supposed errors and wishes at all (no-)costs to
eliminate errors.
As a person concerned with basic map quality, I cannot agree with that
attitude. How can a program decide if it is a real error or a badly coded
definition and then, what correction to bring? (two polygons touching at one
point and forming one region may be coded as an 8 single polygon region if
the node order is not correct) (if two regions overlaps, which border to use
to fit the other?) (If there is a gap between to adjacent regions, is it a
real gap, a lake?, or an error?)
In order to live up to my responsabilities under this light, I have written
an app (reg2plin) that can help in the detection of a large number of
topologically-speaking problematic situations. I have not compared its
output with that of MapCheck, but reg2plin produces graphic output and not
lists. However, in your case, it will not be of any help because it uses the
new graphic motor of version 5.5; I am sorry to have given you false hopes.
But I could resist pointing at these two facts : the basic flaw in MI<5.5
that yields unavoidable EOO, the misconception that topological errors can
be corrected without human intervention.
Incidentally, the appp mentioned here are available on our site (see
signature)
Jacques Paris
============================================================
e-mail [EMAIL PROTECTED]
alternate [EMAIL PROTECTED]
gis activity (MapInfo mainly)
http://www.total.net/~rparis/gisproducts.htm
============================================================
----------------------------------------------------------------------
To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put
"unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]
