Interesting -- this issue didn't arise in my application. Note that 'extract' includes an argument called 'weight', which will apparently return the cell value & the fraction of each cell which is inside the polygon. This should be enough to compute the weighted average in a case like yours. I haven't tried it out, so can't comment on efficiency -- although the help page suggests it can be slow.
Another possibility might be to sample the polygons with regular points using 'spsample' -- and then use the raster values at these point locations (from 'extract') to approximate the weighted average. Cheers, G. On 05/09/13 02:37, Michael Treglia wrote: > Thanks for the suggestion! I ended up using the centroids of my > polygons, and doing the overlay in SAGA GIS. The whole process ran > quickly. > > GD- Ill keep your suggestion in mind for the future. A quick question > though - your method doesn't directly allow computation of weighted > average of pixel values within a polygon, does it? For my situation, > for example, with large raster cells with respect to the polygons, > imperfect alignment means that sometimes a polygon is 1/3 of 1 pixel > value and 2/3 another, so I would want the weighted average of the > pixel values. (I've got what I needed to work for now, but curious > for the future). > > Thanks again, and best regards, > Mike > > > On Fri, Aug 30, 2013 at 11:42 PM, GD <grothe...@gmail.com > <mailto:grothe...@gmail.com>> wrote: > > I had related situation with 85000 non-overlapping polygons, and a > raster > dataset with dimensions ~ 10000x10000. In that case I wanted the full > distribution of pixel values inside each polygon, rather than just > the mean. > > An efficient approach (much faster than extract) was: > > 1) Rasterize the polygons (to a single raster with the same > dimensions as > the input raster dataset). My polygons had a unique ID value that > was burned > into the raster. I did this with 'gdal_rasterize' > > 2) Use the 'crosstab' function in the 'raster' package to > cross-tabulate the > values from the ID raster with the values on the input raster > dataset. This > gives the full range of pixel values associated with each polygon ID. > > 3) From there, it is straightforward to apply whichever function > to the > pixel values for each ID. > > > > > -- > View this message in context: > > http://r-sig-geo.2731867.n2.nabble.com/Extract-function-too-much-data-for-R-tp7584509p7584521.html > Sent from the R-sig-geo mailing list archive at Nabble.com. > > _______________________________________________ > R-sig-Geo mailing list > R-sig-Geo@r-project.org <mailto:R-sig-Geo@r-project.org> > https://stat.ethz.ch/mailman/listinfo/r-sig-geo > > [[alternative HTML version deleted]] _______________________________________________ R-sig-Geo mailing list R-sig-Geo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo
