Hi Etienne, What's your purpose?
First, I will suppose that you try to estimate a key variable "k" with "n" covariables "c". as k = c1+c2+c3+cn To do this, you could treat your raster as a vectorial grid using a primary key and use the package bigmatrix. Bigmemory package give you the opportunity to manage big dataset by writing on the hard disk ( http://www.bigmemory.org/). With this option you could model your key variable and keeping the spatial coordinate with the grid id. If you need to do spatial analysis, I will suggest you to convert your raster in spatial point data frame and convert it back to raster. Hope it will help you. JD 2012/9/24 Etienne B. Racine <[email protected]> > For test purpose, I want to create a large raster within R and write it to > disk. However, I find myself in the situation where I can't write to my > raster since it is too big to manage in memory while not being able to > write to disk since the raster has no values. > > Here are the different strategies I've taken, but all failed : > ``` > r <- raster(matrix(1, nrow = 1e5, ncol = 1e5)) > r <- raster(nrow = 1e5, ncol = 1e5) > r <- setValues(r, 1) > r <- calc(r, function(x) 1, filename = "big_raster") > values(r) <- 1 > writeRaster(r, filename = "big_raster") > ``` > > Any solution ? > > Etienne > > [[alternative HTML version deleted]] > > _______________________________________________ > R-sig-Geo mailing list > [email protected] > https://stat.ethz.ch/mailman/listinfo/r-sig-geo > -- Jean-Daniel Sylvain 215, rue Alleyn Québec, Québec G1K 4T7 Tél:418-523-7253 Courriel:[email protected] [[alternative HTML version deleted]]
_______________________________________________ R-sig-Geo mailing list [email protected] https://stat.ethz.ch/mailman/listinfo/r-sig-geo
