[R-sig-Geo] Raster map from data.frame

[Michael Sumner]

raster::rasterFromXYZ  should do the trick. I always feel a little ill
turning raw tabular data into a raster, it's so easy(!) but represents a
ridiculous amount of redundancy that gets turned into a simple rule - which
is what the data should have been in the first place.

A somewhat more general solution exists in sp, you would do this ("d" is
your data frame with "x", "y", "z" columns):

library(sp)
coordinates(d) <- c("x", "y")
gridded(d) <- TRUE

That would give you a "SpatialPixelsDataFrame", which does two things (that
should be separate IMO):

- allows the grid to be sparse (note that algorithms to detect "regularity"
in any arbitrary set of coordinates is quite hard to achieve, and
impossible in some scenarios - if it's a grid that should be explicit, why
would its extent be defined by populated cells, etc., rant . . .)
- it allows some level of non-regularity, and it  justifies this by storing
the original explicit coordinates so it can restore them if needed

Also note the raster::rasterFromXYZ case needs extra work to apply it to
multiple-attributed grids (no matter if you have one "z" column or any
number of them the sp case works fine). You can also ingest any sp grid
into raster() or raster::stack() or brick(). If rgdal is present you can
writeRaster() to a .tif as well.

If your data really are not regular (why would you store all those
coordinates?), then you might
- rasterize from irregular data (or explore other model solutions from
packages like gstat)
- use sp::points2grid to control the threshold of "regularity" (really this
is for getting around numeric fuzz, but can be used)

(note also that rasterFromXYZ has a digits argument, similar to
points2grid's eps)

HTH





On Thu, 9 Apr 2015 at 22:44 Maurizio Marchi <mauriziomarchi85 at gmail.com>
wrote:

> Hi everybody, I need help with a big dataset that I need to transform in a
> group of raster files. I have a big .csv (2 GB with 5 millions rows and 90
> columns) which I want to transform in a group of raster images. The First
> two culms represent the geographical coordinates while the others 88
> columns represent the values I want to give to each cell of each raster
> (temperatures, precipitations etc...). The database is compiled only with
> the cells that will have a value (NA cells were not calculated).  I often
> used the data.frame2ascii function of SDMTools package but in this case the
> function doesn't works (too many rows maybe?).
> Any suggestion? I prefer the .asc format but even GeoTiff or other formats
> would be ok.
>
> Manu thanks,
> Maurizio
>
>
>
> --
> Maurizio Marchi
> Calenzano (FI) - Italy
> ID Skype: maurizioxyz
> Ubuntu 14.04 LTS
> linux user 552742
>
>         [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

	[[alternative HTML version deleted]]