Take the 2-minute tour ×
Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It's 100% free, no registration required.

I would like to create smooth/distinct polygon zones from raster data that was interpolated from sparse point data. Is the only way to do this is to have a very small output cell size during the interpolation process? I don't like the look of the raster data when I zoom in and see the grids from the raster data. Is there a way I can eliminate these raster grids and have truly distinct polygon zones? In addition, it would be nice if I can change the number of classes that make up the zones easily by simply changing the symbology and changing the number of classes used or even the classification method used.

share|improve this question
    
It's be nice to have a more specific description of what your raster now looks like and what you'd like it to look like. What specifically bothers you about it? The pixelated edges? –  Patrick Sep 15 '11 at 6:25
    
Yes, the pixelated edges don't look good. And on top of that I don't like the smaller inclusions since they are too small of an area to be relevant. As for what the raster looks like, it's basically 20 meter grids with several areas of single grids having different colors due to different values. I would like to make it where say there would have to be > 10 adjacent grids of the same range values or it just ignores those values. –  wilbev Sep 16 '11 at 1:52

2 Answers 2

As for reclassification, doesn't that depend on how you create the interpolated raster in the first place? I was under the impression that the interpolation function gives you several distinct classes.

This requirement for easy reclassification does seem to conflict somewhat with your need for eliminating the raster grid. Normally I'd recommend you vectorize your raster data (to get distinct areas) and generalize the resulting vector (to get rid of the raster grid artifacts), but you obviously can't reclass a vector.

The only way I can think of to harmonize these requirements is to build an analysis model with some adjustable parameters, and run that every time you want a new classification. In that case, depending on how accurate you want it to be, you might be able to get away with a relatively low-resolution interpolation raster, since you're going to vectorize/generalize the result anyway.

I'm assuming that processing time matters - if it's not critical, then maybe just try a higher resolution raster from the get-go.

share|improve this answer
    
Thanks for the comments R Thiede. It's surprising there isn't a better solution for this. I don't really like the idea I would have to reclassify when I ever need to change the number of classes. A legend change would be much better like I can do with a raster. As I said in my comments to Patrick, I think this would all work better if I could eliminate the smaller polygon zones or even raster zones during the interpolation. –  wilbev Sep 16 '11 at 1:48
    
I don't have Arc right now, I'm going off of what I can remember from using these functions a few years ago. So I'm not a hundred percent sure if you can eliminate small areas during interpolation somehow. Your best bet is to look for another function to do that. Reclassifying has to be done at raster level tho, as it's impossible to do with a vector. Also you may want to polygonize and then re-assign polygons with small areas to the surrounding class, if appearance is more important than accuracy. Sorry I can't be more specific, as I said I don't have Arc in front of me right now. –  R Thiede Sep 16 '11 at 13:09

The standard way to do this in ArcGis using Spatial Analyst, is to:

  1. Reclassify your data, grouping ranges of values into categories, using Spatial Analyst Tools > Reclass > Reclassify.
  2. Convert the raster to a polygon, using Conversion Tools > From Raster > Raster to Polygon. This tool will allow you to smooth the edges of the polygons, removing the pixelated look.

If you want to be able to dynamically change the classification, you probably should leave the data as a raster and change the classification in the symbology tab of the layer properties.

share|improve this answer
    
Patrick, thanks for the tips. When I go from Raster to Polygon, it doesn't seem to smooth the polygon zones as much as I would like it to. I think if I could eliminate the smaller polygon zones it would work better. Do you know of any way I could make so the zones are at least say over 0.5 acres? It would be better if I could do this during the interpolation so I could change the classification of it on the symbology tab –  wilbev Sep 16 '11 at 1:44

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.