Building multi-scale rectangular grids based on raster zonal statistics.

Dear all,

I am using some algorithm of my own for building multi-scale rectangular grids with a refinement depth based on zonal statistics performed on a raster (e.g. all grid cells whose sum of pixels values is above a given threshold are refined), as illustrated below

I was wondering if any of you knew a powerful library or algorithm for doing this kind of operations in MATLAB (I am currently doing this in both MATLAB and ArcGIS/Python for comparing), because my algorithm is starting to be slow with e.g. 8000x4000 rasters.

Thank you and best regards,

Cedric

Answer 1

[Jan]

I've used a leap frog algorithm, see e.g. "Structured adaptive mesh refinement using leapfrog time integration on a staggered grid for ocean models" Ocean Modelling 9 (2005) 283–304, doi:10.1016/j.ocemod.2004.07.002 . But unfortunately this was an experimental program only, which had an advantage compared to using a fine mesh only, when the coarse mesh avoid disk swapping.

I assume you know adaptmesh already, but perhaps this tip it might be useful other readers.

[Cedric]

Hi Jan, thank you for your answer and sorry for the belated reply! The issue with adaptmesh and related functions is that it is difficult to use them for generating rectangular multi-scale grids like mine, but thank you for your reference; I didn't know this article actually and I am placing it right now in the pipe of the articles that I have to read during the upcoming week!

Cheers,

Cedric