Hi Guys!
I guess that honestly I get serious problems with the raster package in R and their projections tools.
I got information about SST from Nectdf files in area along south america.
I need transform, the raster obtained, to Geotiff or ascii files. I transformed the files in both format (to get the ascii file I resample the original raster). So far, so good I got the files wih identical size (rows, column, pixel size) that my asci variables obtained with ArcGis.
However, when I run the code to project the files and try to map with any tool (ggmap, wrld map) I obtained always the same results .
The Rplot image is my variable obtained from ArcGis as you see it's not adjusted with wrld map.
The Rplot01 is much more my concern because it is the file obtained from netcdf file !!!. If you can't see the file projected it's the verysmall green square next to african coast (!!!!))
I'm so frustrated because I've tried with different projections, file types and always I got the same results. If I load the file in arcgis or Qgis, it's appears next to Africa too!!!
Any comment or help will be so appreciated
This is part of my piece of code
SSTImage
Hi Ricardo,
You may have a look at older RG questions, or threads in the R-SIG-GEO mailing list: see the links below.
Difficult to know what went wrong in your case without the Ascii, but my guess would be that the original netCDF data was not projected, but in Lat/Lon (this might explain why some stuff appears in the Gulf of Guinea in your second plot)
Modern GISes such as ArcGIS have the ability to project vector and raster layers 'on the fly', so discrepancies in projections may not appear as a problem until you're in R, where things have to be more coherent.
Why no start from scratch in R, creating your layers directly from the netCDF file? (the command raster(xx.nc) should work - or brick() if needed). To reproject to your adhoc conic projection, you would then rely on gdalwarp (packages rgdal and gdalUtils required).
regards,
http://r-sig-geo.2731867.n2.nabble.com/Problem-with-raster-package-and-nc4-file-td7588429.html
https://www.researchgate.net/post/How_to_convert_a_NetCDF4_file_to_GeoTIFF_using_R2
Hi Baptiste,
First of all many thanks for your time to respond and links, I'm reviewing. I think that I must explained better the first part of my problem. I got the information from netcdf file, using
ncatt_get(data,"time","units") ## time in "days since 1800-1-1 00:00:00"
ncatt_get(data,"sst","units") ## sst in degC
nc_close(data)
Then I got a period of time with the brick command and then, I've obtained only one file using the function calc.
My problem it's produce to export that file to my ArcGis system (geotiff, asci, grid whetever).
In addition, I've tried to convert the file in a Geotiff file and I got the same problem, evenmore, the size of the tif file is extremely small (34kb). I'm not sure if I'm doing something wrong, but I guess looks like a snapshot more than a geotiff file properly. What do you think?
Cheers and many thanks again!
Ricardo
Hi Ricardo,
the pictures and description you have posted clearly point out the problem of projections as you've already known :-) When I checked the code snippet I noticed that you are using crs function while you might wanted to use CRS instead. Both functions exist in R (crs in raster / CRS in rgdal) but they have slightly different usage - please see their help for details. So if you adjust the code it might work (or other problems may appear :-) I agree that the doubling of functions names in different packages makes life (and programming) much more fun :-D and agree with Paulo that with some data might be useful for such a fine tuning.
Best
Lukas
You can try this:
conicprj
Hi Ricardo,
In order to get the fastest and the most efficient help from RG users, a good practice (standard in places such as Stack exchange or the R mailing lists) is to provide a reproducible example, with an access to data and your code.
Here is the workflow I would suggest:
setwd("C:/DATA/TEMP")
library(raster)
# We start by importing a SST dataset.
# Mine was downloaded here: http://esrl.noaa.gov/psd/thredds/catalog/Datasets/noaa.oisst.v2.highres/catalog.html?dataset=Datasets/noaa.oisst.v2.highres/sst.day.mean.1981.v2.nc
# The 'brick' command, from the package 'raster', imports your netCDF as a 'RasterBrick'
a_raster_brick
Thank you very much for your help guys and I'll consider your advice for my next post Basptiste!!
Cheers!!!
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping
- Maps