I am studying migration ecology of sea ducks on the Atlantic coast of North America.
In my analysis, I require substantial environmental data over a large geographic range for the past 15 years and have had great success in using the ICOADS database for most of my desired model parameters. However, they do not track precipitation and left me needing a different source for it. NARR is the only source with a narrow enough scale to be of real use for my purposes but I am finding it much less simplistic than ICOADS.
NARR uses a Lambert Conformal Conic projection and I am having some difficulty in reprojecting it into long-lat form. I have consulted a few other similar forum posts, such as R: How to get latitudes and longitudes from a RasterLayer? , but am still confused regarding what the suggestions really mean and if I am applying them to my own data correctly.
Following the above link, I used this coding to raster and reproject my precipitation data file (here is the link to file apcpc.2000.nc in case anyone would like to play with the data themselves: ftp://ftp.cdc.noaa.gov/Datasets/NARR/Dailies/monolevel/):
Loading packages and rastering file
library(raster) library(rgdal) r<-raster(file.choose()) #I choose the apcp.2000.nc file
Metadata of raster
> r class : RasterLayer band : 1 (of 366 bands) dimensions : 277, 349, 96673 (nrow, ncol, ncell) resolution : 32463, 32463 (x, y) extent : -16231.5, 11313356, -16231.5, 8976020 (xmin, xmax, ymin, ymax) coord. ref. : +proj=lcc +lon_0=-107 +lat_0=50 +x_0=5632642.22547 +y_0=4612545.65137 +lat_1=50 +lat_2=50 +ellps=WGS84 data source : C:\Users\k0i91\Desktop\MSc Data\NewPrec\apcp.2000.nc names : Daily.accumulated.total.precipitation.at.Surface z-value : 2000-01-01 00:35:47 zvar : apcp
Current projection of the raster
crs(r) <- "+proj=lcc +lon_0=-107 +lat_0=50 +x_0=5632642.22547 +y_0=4612545.65137 +lat_1=50 +lat_2=50 +ellps=WGS84" #This is the coord. ref. shown in the metadata of r above
Convert raster to SpatialPointsDataFrame
r.pts <- rasterToPoints(r, spatial=TRUE) proj4string(r.pts)
reproject sp object to long lat
geo.prj <- "+proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0" r.pts <- spTransform(r.pts, CRS(geo.prj)) proj4string(r.pts)
Assign coordinates to @data slot, display head and summary
r.pts@data <- data.frame(r.pts@data, long=coordinates(r.pts[,1],lat=coordinates(r.pts)[,2]) > head(r.pts@data) Daily.accumulated.total.precipitation.at.Surface long lat 1 4.883111e-05 149.2377 48.19028 2 2.200049e+00 149.3178 48.47717 3 5.300049e+00 149.3988 48.76403 4 6.300049e+00 149.4807 49.05085 5 6.400049e+00 149.5637 49.33761 6 4.800049e+00 149.6476 49.62432 > summary(r.pts@data) Daily.accumulated.total.precipitation.at.Surface long lat Min. : 0.00005 Min. :-180.00 Min. : 0.847 1st Qu.: 0.00005 1st Qu.:-135.33 1st Qu.:23.196 Median : 0.20005 Median : -98.59 Median :39.988 Mean : 1.75934 Mean : -86.69 Mean :40.386 3rd Qu.: 1.20005 3rd Qu.: -64.82 3rd Qu.:56.015 Max. :44.90005 Max. : 179.99 Max. :85.260
It appears to work well enough but I second guess on what I am supposed to enter as the initial CRS here. It just makes intuitive sense to enter the raster's listed coordinate reference but the example in the post I have consulted does not do this (they use: "+proj=lcc +lat_1=48 +lat_2=33 +lon_0=-100 +ellps=WGS84"). Additionally, NARR states the boundaries of coverage only extend to roughly 145 W and E and just to 45-50 N; however, I seem to have values for the entire Northern Hemisphere. This could easily be caused by updated coverage since that was stated several years ago but serves to fuel my concern I erred somewhere.
Also, rastering does not seem to provide the "time" variable anywhere. Is there a way to extract this as well? As there are 366 days in this file, it would be annoying but not difficult to add them in later based on repeating coordinate pairs.
I am hoping someone has experience with the NARR database (or just converting lcc to long-lat) and can explain if and how I did this wrong, then steer me down the correct path?
***UPDATE**** Plotting r.pts Longitude and Latitude reveals the coding works just fine for the intended reprojection. But I'll leave the post open in case someone can suggest something for the "time" variable.