3

Let's say I have a vegetation raster that indicates the density of forest all over the world. I also have a shapefile that contains the admin boundaries of each country in the world. The easy part is to make sure the grid cells are within the country boundary, but what I want to do afterwards is to create a separate polygon boundary that will surround the region in question. It is best to do it dynamically based on a certain density threshold, that could vary from country to country. I haven't exactly figured out how to handle that yet, but once I do, how would I go about encasing that data within a polygon that can be saved as another shapefile. I have attached an image of what I want the result to look like. Polygon

So after a threshold is determined, a polygon will be drawn around the area as tight to the boundary as possible, excuse the freehand above, and then output it as a new shapefile. Is this possible and any ideas on how it can be done? Would using Model Builder be feasible to create a tool that would take the density shapefile, boundary shapefile, and threshold value as inputs to produce a polygon?

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
twistr25
  • 343
  • 4
  • 15
  • You might find these discussions useful: http://gis.stackexchange.com/questions/8/how-do-i-create-a-convex-hull-in-arcgis-desktop – Branco Aug 05 '14 at 17:52
  • Indeed they are, turns out though, I have a raster file and not a shapefile like I thought, so that opens a whole other can of worms. OP fixed to reflect this. – twistr25 Aug 05 '14 at 19:09
  • This question seems to be almost identical to the one at http://gis.stackexchange.com/questions/102468/how-to-group-individual-groups-of-pixels-into-a-single-class. How, if at all, does it differ? – whuber Aug 05 '14 at 19:21
  • Looks pretty similar. I'll try some of those techniques and see what I can come up with. – twistr25 Aug 06 '14 at 12:19
  • I'm going to be honest @whuber, that is beyond my capabilities most of what you discussed in that post. I have resorted to using focal statistics to create a point shapefile of the region I want, and it seems like it should be really easy to trace an outline of the points, but I am having a real hard time figuring out how to do that. – twistr25 Aug 06 '14 at 15:45
  • I doubt it's beyond your abilities, because all it requires is to compare the output of a kernel density calculation to a threshold. The result is a binary grid that represents the desired polygon--and can readily be converted to a vector polygon feature if desired. You can control the shape and appearance of the polygon through the KDE parameters (kernel shape and width) and the choice of threshold. – whuber Aug 06 '14 at 15:47
  • Ok, let's say I figure that out, I guess I forgot to add my end game in the OP. I need to extract the lat/lon for the polygon area in order to ingest it into a forecast model. Is that in any way possible? – twistr25 Aug 06 '14 at 16:28
  • How are you going to get lat/lon for a polygon? Coordinates for every vertex? Easiest way I can think of is to create centroids from your polygons and then add X/Y coordinates to your attribute table. That is easily done in Model Builder or Python. – Baltok Sep 21 '15 at 13:22

1 Answers1

1

Is it possible to Convert Raster to Polygon, then use the polygons in one of the tools referenced by Branco's comment?

Feature Outline Masks

Minimum Bounding Geometry

Barrett
  • 3,069
  • 1
  • 17
  • 31
  • I tried this, but my values are fairly small and it requires the Raster values to be integers. I'm sure there is a way around it, like multiplying by a large number then re-dividing later, but when I first did it, I got really large grid cells that were no longer representative of the area I was working in. But I am exploring this option as well. – twistr25 Aug 06 '14 at 12:21