[R-sig-Geo] DEMs

[Tomislav Hengl]

Hi Abe,

Did you try using the ETOPO1? This (0.1 degree) DEM can be well loaded to R:

> URL <- "http://spatial-analyst.net/worldmaps/"
# list of maps:
> map.list <- c("globedem", "chlo08", "dcoast")
# download the zipped maps one by one:
> for(i in 1:length(map.list)) {
>  download.file(paste(URL, map.list[i], ".zip", sep=""),
+   destfile=paste(getwd(), "/", map.list[i], ".zip", sep=""))
>  unzip(zipfile=paste(map.list[i], ".zip", sep=""), exdir=getwd())
>  unlink(paste(map.list[i], ".zip", sep=""))
> }
# read maps to R:
> worldmaps <- readGDAL(paste(map.list[i], ".asc", sep=""))
# fix the layer name:
> names(worldmaps)[1] <- map.list[1]
> for(i in map.list[-1]) {
> worldmaps at data[map.list[i]] <- readGDAL(paste(map.list[i], ".asc",
sep=""))$band1
> }
> proj4string(meuse.grid) <- CRS("+proj=longlat +ellps=WGS84")

I do not think that it will be easy to generate kriging with any global
dataset that is <5 km resolution. You could try using the kriging in SAGA
(it can handle maps up to 2 GB in size; it is exp(2) times faster than R).
Here is an example:

http://spatial-analyst.net/wiki/index.php?title=Interpolation_of_ISRIC-WISE_international_soil_profile_data

T. Hengl

PS: Which variable are you interpolating?

> Hello All,
>
> My questions are regarding the best way to obtain DEMs for the entire
> globe.  I am trying to krige a dataset of global scope.  The dataset has
> points at all latitude/longitude intersects and so there are 64,800 sample
> points.  The problem that I am having is that the DEMs I have been using
> are
> too large (memory) to be used as R variables.  I also have had trouble
> finding a way to download DEMs for the entire globe.
>
> 1.  Given the size of my sample dataset: What is the maximum resolution of
> DEM that I should be using?  Do I only need values at prediction points or
> will higher res yield higher accuracy?
>
> 2.  Is there a way to programmatically access DEMs using GDAL or R in
> general?  ie. internally or through integration with McIDAS, NetCDF, etc.
>
> 3. Do I need to break the DEMs and sample dataset into smaller tiles to
> avoid the memory issues?  If so how do I do that in R / GDAL?
>
> Thanks,
> Abe
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>