AnsweredAssumed Answered

calculate zonal histogram for each pixel in raster

Question asked by co.bzk on Feb 3, 2015
Latest reply on Feb 24, 2015 by co.bzk

Hello to everyone,


I am trying to create a habitat map based on a land cover map, and as it seems a big fish for me I prefer to ask for advice beforehand, and I thank very mcuh anyone who'd be willing to help !


I have a land use map at 15x15m with ~ 10 land uses, and a habitat classification based on land cover distribution around a point.

ex: Habitat "3" = Mosaic cropland (50-70%) / vegetation (grassland/shrubland/forest) (20-50%)


I would like to perform the following steps, for each pixel in the land use map:

1) get a pixel count of each of the land uses within a 150m radius

2) confront the land use distribution with my habitat classification directly in arcpy

3) attribute the habitat code value to the point, and later convert to raster.




And here are the points about which I'm wondering:

1) should I work in raster format and count pixels? or in shapefile format and calculate areas? I think raster.

2) should I implement a moving window, or convert each point in my shapefile to a 150m-wide circle? I don't even know whether zonal histogram would give me the correct result with that many overlapping features. (I've also thought of building several shapefiles of touching but not overlapping circles, but that's for later I think)

3) should I try and use multi-processing? I'm fairly new to Arcpy, so this might be tricky.

4) is it possible to confront tables in arcpy like in R or Matlab or should I export the distribution table to another software to run these calculations?

5) in a more general context, how to determine what is a reasonable amount of calculation? I have a sense 14 million iteration sounds stupidly high, but I couldn't get more precise...


Thanks in advance for any help!