RGB Raster Image of Regional Percentages of Food Scarcity

I am completing an analysis project for my ArcGIS class. I am analyzing the correlation between indigenous populations and food security in Guatemala. Since there is not a lot of information out there on this topic, I had to georeference .jpg maps I found online. I now have two RGB raster datasets: one shows the distribution of indigenous populations, and the other shows regional percentages of food scarcity. I would like to display the indigenous population raster layer as Quantities laid over the regional areas of food scarcity, therefore I have to convert it to vector data. I am having trouble finding a tool that can do this.

  • The image appears to be mostly blue/green. What does the blue band look like? Being a JPG you're not going to get clean boundaries unless you heads-up digitize over the lines. How important is it to get this accurate? Dec 8 '14 at 0:52
  • There are a lot of ways to approach this, and I'm not clear whether you're just trying to display or actually do analysis. Layers have a transparency setting on the Display tab, though that likely won't cleanly get you what you want. To do analysis, staying in raster might be an option if you have Spatial Analyst. To get from raster to vector you need... Raster to Polygon, but again you may not like the results. Are the referenced maps from the same source (ie, do they overlay nicely, same resolution)?
    – Chris W
    Dec 8 '14 at 1:08
  • Possibly the best approach is to contact the agency that published the map and ask (nicely) if they still have the data. It's going to be way more accurate than raster to polygon, 3-band to 1 classified and faster than heads-up digitizing... but if you've got students that want extra credit there's a bread-and-butter GIS task you can hand on to them. Dec 8 '14 at 1:10
  • 1
    Since I see you have municipality borders as polygons, you might add two attributes to that layer, use feature to point constrained to inside, extract values to points (once for each raster), then join the points to the polygons and write out a new file so you have the attributes saved. Note that with RGB you would likely have to do cleanup since green = 245 or 242 is possible though they should be the same color. You could then add the layer to your map twice and symbolize on the two different attributes for each. This assumes your two maps are already choropleths like you show.
    – Chris W
    Dec 8 '14 at 1:33

You have the municipality boundaries as a shapefile, and your jpeg has "homogeneous" values inside each municipality. Therefore, the best method to get your values is tansfer the values of eac band. It is easy if you have spatial analyst licence:

1) feature to point with the "inside" option to build the centroid (or add XY fields and make XY event layer if you don't have the licence)

2) extract multiple value to point

3) join the points' table to the polygons' table and you have the RGB attributes for each boundary.

Now it depends how your map was coded: either each banb has a meaningfull value and you are done, or this was a colormap and you need to find the relationship between the RGB values and the actual values (need to have information such as the legend to infer this)

  • This worked well! Thank you so much for your help! Dec 8 '14 at 23:34

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.