Ask A Cartographer

What drainage generalization methods are available in ArcGIS 9.2?

February 21 2008 | 1 comment
Categories: ArcGIS Methods, Data Modeling

What drainage generalization methods are available in ArcGIS 9.2?

Mapping Center Answer:

So, there's really not a short answer to this question. To assume that there are "methods" is an oversimplification of the problem. There are two contexts that I can think of for approaching this problem; one is raster based and the other is vector based. I'll try to give you a start in both:

Raster: The Spatial Analyst Extension arguably has the most sophisticated tools for both analyzing and generalizing drainage. The idea is that you can extract drainage from your elevation model, generalize that elevation model, and extract drainage from the generalized model. (that's a gross oversimplification in that I am not going to specifically touch on the issues involved).


Vector: This is a pretty complex process and most hydro vector data is not captured with the necessary information or captured into a data model that readily supports genearlization. We tend to aggregate hydro vector geometries based on either semantic or network requirements; neither are sufficient (by themselves or together) to support generalization. The initial problem is to select the polygons and lines that will be used to derive the smaller scale data. The NHD team at the USGS has spent a good bit of time on this and has made some progress with automating stream selection within a stream network and doing so based on catchment areas. None of that work deals with polygons, just the NHD center of channel lines. ESRI is working with the USGS on methods for selecting and simplifying polygons, but none of that work is published to date. The gist of the work is in removing the insignificant polygons based on their size and shape type and then when an eliminated polygon overlays a channel line, convert that channel line to the appropriate type of stream line. The remaining polygons are generalized by converting them to lines, intersecting (breaking) those outlines where they intersect streamlines to preserve the junctions and simplifying the lines using the Simplify Line tool with the Bend Simplify Method. The details of this method are "generically" discussed in a Mapping Center blog posting called, Simplifing lines and polygons that intersect with other lines and polygons. The data and method were developed while working on this as yet unpublished research.


So, as you can see that just yet this is currently more of a field of study rather than a method.

Buffer/Centerline routine (using ET Geowizards) posted by Craig McCabe on Feb 21 2008 2:19PM
I was recently up against the same issue with generalizing a detailed 1:24k fault network and representing it at 1:250k, which is somewhat similar to your drainage network problem. The process involved some trial and error and a plug-in, but this is the vector-based approach I ended up taking:

1) Buffer the line network to create a new polygon layer (the correct buffer distance took some experimentation and you may need to break up the data and apply different buffer values in different locations, depending on whether neighboring lines should coalesce or not)
2) Use the ET Geowizards plug-in to Create Centerlines from your polygon layer(s)

That was it. Depending on how your drainage network looks, this may or may not be a viable option, but it's worth a shot. My first attempt was to use Collapse Dual Lines to Centerlines, but this never got me a satisfactory result (I got lots of abrupt right-angles as soon as the neighboring branches exceeded the specified separation distance).

Good luck!

If you would like to post a comment, please login.

Contact Us | Legal | Privacy |