I need to calculate centroids for each feature in a polygon layer, weighted by the values of a raster. So the centroids are shifted towards higher raster values.
I found a way, but it is complicated. Is there an existing tool which can do this?
-
1Why not just iterate through each polygon, mask to polygon i, calculate a probability density function (PDF) and then take the mode. This represent a spatial process in a more relevant way.– Jeffrey EvansCommented Oct 28, 2021 at 13:55
-
That might work, I will try it– BeraCommented Oct 28, 2021 at 13:57
-
I have to admit that I have R or Python in mind where a vector of raster values can be modeled and indexed. In this way a pdf can be derived (eg, density, ks), the index of the mode returned along with the corresponding [x,y] coordinates. Not sure how realistic this is in QGIS but, there is likely a way for somebody savvy with the software (which is not me).– Jeffrey EvansCommented Oct 28, 2021 at 14:03
Add a comment
|
1 Answer
I've done similar approaches with vectors to get population based centroids. I don't know any existing tool that does this but it would be neat.
Here is another approach, it is also complicated but maybe a bit more precise than grid + variable buffer.
It is a vector based approach, basically weighting coordinates by pixel value. If there was a way to create a raster where pixel values were x and y coordinates without point conversion, would be easier.
And here is the result
