[R-sig-Geo] Convert rasters to data frame with time stamp

[Vijay Lulla]

>From the help page of `getValues`: "The values returned for a
RasterStack or RasterBrick are always a matrix, with the rows
representing cells, and the columns representing layers"  ....so all
you have to do is transpose the matrix from getValues, cbind the date
column, do your analysis.  To go back to raster again just drop the
names/date column, convert to matrix, transpose, and do setValues.

So for your example (untested...will need checking):
R> s_t <- t(s)
R> df_s_t <- cbind(names(s),as.data.frame(s_t))
R> # do your analysis
R> # To do the reverse
R> m <- as.matrix(df_s_t[,-1])
R> mt <- t(m)
R> setValues(s,mt)

HTH,
Vijay.

On Fri, Oct 16, 2015 at 4:39 PM, Thiago V. dos Santos
<thi_veloso at yahoo.com.br> wrote:
> Dear list,
>
> Generally speaking, I love R. However, one of the things I like least in R is the need to interchange between the various data formats required by different packages.
>
> I am trying to apply a bias-correction function on some gridded climate data. The qmap package has functions to perform bias correction on climate data, but the problem I am grasping with is that it requires data to be organized as data.frames:
>
>
> library(qmap)
> data(obsprecip)
> data(modprecip)
>
> #Fit a quantile mapping function to observed and modeled data
> qm.fit <- fitQmap(obsprecip,modprecip,
>               method="QUANT",qstep=0.01)
>
> #Perform bias correction on modeled data
> qm <- doQmap(modprecip, qm.fit, type="tricub")
>
>
> And that's all. But notice that both observed and modeled data in this example are data frames for different locations (Moss, Geiranger and Barkestad):
>
>> head(obsprecip)
>          MOSS GEIRANGER BARKESTAD
> 1-1-1961  0.1         0         0
> 2-1-1961  0.2         0         0
> 3-1-1961  0.9         0         0
> 4-1-1961 10.6         0         0
> 5-1-1961  1.5         0         0
> 6-1-1961  1.2         0         2
>
>> head(modprecip)
>            MOSS GEIRANGER BARKESTAD
> 2-1-1961  2.283    0.0000  3.177000
> 3-1-1961  2.443   10.8600  1.719000
> 4-1-1961  3.099   12.7300  6.636000
> 5-1-1961  0.000    9.7720  9.676000
> 6-1-1961  0.140    0.6448  7.110000
> 7-1-1961 13.470    3.3570  0.001107
>
>
>
> Now, let's back to my problem. I have monthly precip data to which I want to apply the same function above, but my data is gridded:
>
> library(raster)
>
> #Create a rasterStack similar to my data - same dimensions and layer names
> r <- raster(ncol=60, nrow=60)
> s <- stack(lapply(1:408, function(x) setValues(r, runif(ncell(r)))))
> names(s) <- paste0('X', seq(as.Date("1980/1/1"), by = "month", length.out = 408))
> s
>
>
> Therefore, I need to load data as rasters and iterate through all individual gridcells to create a data frame containing:
>
>
> date1, cell1, cell2, cell3, ..., cell3600
> date2, cell1, cell2, cell3, ..., cell3600
> date3, cell1, cell2, cell3, ..., cell3600...
> date408, cell1, cell2, cell3, ..., cell3600
>
> then apply the fit function and finally convert the data back to a raster.
>
>
> Any ideas on how to efficiently convert rasters to data frames containing their time stamp and then back to a raster again??
>
> Any hint is much appreciated.
> Greetings,
> -- Thiago V. dos Santos
>
> PhD student
> Land and Atmospheric Science
> University of Minnesota
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo