1

I need to produce a shapefile from a raster. I need to be able to specify either an interval, or manually define the intervals. The input file is a GeoTIFF. I would like to use GDAL and/or Python.

Example raster:

enter image description here

I tried using gdal_contour, which at first glance would have been perfect. However, I want the result to have the same "pixel" definition - that is, the "contours" should maintain the block appearance of the pixels at their extents. For this reason, I can't use gdal_contour, as the results look like this:

enter image description here

Is there another way to do this? My output should be a shapefile that looks like a grid, with only pixels in the same interval connected to each other.

  • You (may) have two questions here. The first is about reclassifying your raster (specification of intervals) for which you can see gis.stackexchange.com/questions/116473 Contour lines is an interpolation of the point values, not a direct vectorization. For that you can see gis.stackexchange.com/questions/128139 which discusses why it doesn't make sense to directly vectorize a float. However if that's really what you want to do, it can be done. Are you wanting lines or polygons for the vector geometry? – Chris W Jun 14 '15 at 19:08
  • 1
    rasterToContours in the R package raster returns cell edges in a SpatialLinesDataFrame. – mdsumner Jun 14 '15 at 21:10
  • Hey @ChrisW - thanks for the response. My wording probably wasn't great. I want to have the option to specify intervals for the "contours" in my output. So, for instance, 10-30 should be a contour, 31-40 should be one, etc. gdal_contour can do this with the -fl argument. Just want to retain that functionality. Anyway, the more important part: yes, I do want to directly polygonize the raster. It's not really an elevation, it represents a specific set of data that must go into the final system in this format. I'll take a look at your second link, looks like it might work. Thanks! – Dan Jun 14 '15 at 23:52
1

Creating contours is an interpolation, which is why your output looks like it does.

You need to reclassify using gdal_calc: http://www.gdal.org/gdal_calc.html

But there is some good information here relating to reclassifying with either gdal or python: Reclassify rasters using GDAL and Python

And then convert to vector using gdal_polygonize: http://www.gdal.org/gdal_polygonize.html

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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