# [postgis-users] Simplify a perfect (hierarchical) partition of polygons (ST_SimplifyPreserveTopology and friends)

Martin Tomko tomkom at unimelb.edu.au
Wed Mar 14 17:45:43 PDT 2012

```Hi Sandro, William, all,
Thank you very much for the responses.

Sandro, this looks hopeful. I have checked and indeed the
addTopologyColumn method contains the integer specifying a child layer. I
"assume" that this is taken from the topology.layer table where each
topological layer is registered, right?

The "algorithm" noted by William is something of the type I had in mind,
but I think it still does not cater for the problem of reduced set of
controlled nodes in higher levels of hierarchy:
The algorithm was
1. Simplify the smallest (admin2) units using the Simplify Polygon tool,
asking it to resolve topological errors.
2. Dissolve the generalized polygons on their parent id
3. Repeat 2 until I arrived at admin0

I am afraid that this will not cater for some problems (referring to
https://martintomko.files.wordpress.com/2012/03/hiera_generalization.png
now)
In this figure, the thin black lines are say suburbs, and the ticker
purple lines countiesŠ. Even when generalising the county boundaries, I
would want not only the "real" nodes of that hierarchical level (the blue
circles) to be respected, but also those of the previous level (red
circles). Of course, as we go higher, this reduces the generalization /
compression / but that is inevitable.

Sandro, you say:
Quote:
"Yes, it probably don't make sense to simplify single TopoGeoms in turn if
you know you're going to simplify the whole topology. In that case you
would
just make a copy of the topology and then simplify all edges on the copy.
Existing TopoGeometry objects (appropriately re-bound to the new topology)
would magically convert to a simplified version of simple geometries. Both
simple and hierarchical ones."

Is this a "wished for" functionality, or is this doable? I have little
experience with the topoGeom, a code snippet would make me very happy ;)

Thanks,
Martin

P.S.: I do not want to get into the flame of public institutions not
supporting pledges etc. Sometimes it is not for not willing to, but for
administrative reasons. While I am at an academic institution and I think
we could get access to "the dominant GIS system", I do not have it handy,
and we are an explicitly OS project. We are willing to contribute back
(also in due course to other projects, such as Geotools), but we cannot
contribute \$, only code. For those interested, look up www.aurin.org.au

Martin Tomko

```