[R-sig-Geo] downscaling grid data; alternative for interp()?

[Klaus Vormoor]

Hi,

I am struggling with this quite a while already, and I haven't found a good
solution so far.
I want to disaggregate precipitation data. For this I have a grid with daily
precipitation in 1x1 km resolution, and additionally 24 grids with 10x10 km
resolution giving percentage charge of hourly precipitation for the same day
as factors (range 0 to 1). The first grid is in UTM coordinates and the
other 24 are in longlat coordinates. Now, I want to transfer the factors
from the 10x10 km grids to grids of the same size as the 1x1 km daily
precipitation grid. What I have done so far is: 

1. Generate SpatialPoints with the longlat coordinates for the 10x10 km grid
(comming as netCDF).
2. Convert longlat coordinates to UTM coordinates (spTransform), adding the
data values to a SpatialPointDataFrame
3. Interpolate data from the converted SpatialPpointDataFrame to a grid of
the same size and extention as my 1x1 km grid by using the function interp()
from the akima package.  

However, instead of interpolating I'd rather like to use a kind of nearest
neigbour classification because I want to avoid "smoothing" of my factors.
Furthermore, interp() is relatively time consuming. Since I have to repeat
this procedure for many days, it would be nice to have a quicker
alternative. Does anyone have an idea?

Any hint is much appreciated.

Klaus  



--
View this message in context: http://r-sig-geo.2731867.n2.nabble.com/downscaling-grid-data-alternative-for-interp-tp6819843p6819843.html
Sent from the R-sig-geo mailing list archive at Nabble.com.