by GDAL: gdal_translate -of GTiff TEST.nc TEST.tiff

I figured out the answer some time ago but I hope this will help other people that are dipping their toes into the NetCDF file format.

I found that the easiest way to create a new NetCDF file containing several GeoTIFF files is using GDAL and then using NCO and CDO tools (Provided by unidata) to add the appropriate metadata.

First you convert the GeoTIFF´s to NetCDF using:This creates a netCDF in NC4 format (which is the newest version at this time) with some basic metadata like coordinate system, pixel size etc.

Then I rename the variable information to what I want it to be (In my case SST) with the appropriately registered long and short name From the unidata convention page):

These links can be helpful:

https://www.researchgate.net/post/How_to_convert_a_NetCDF4_file_to_GeoTIFF_using_R2

https://gis.stackexchange.com/questions/323317/converting-netcdf-dataset-array-to-geotiff-using-rasterio-python

https://www.mathworks.com/matlabcentral/answers/353583-netcdf-into-geotiff-tiff-file-conversion

netCDF files are generally gridded datasets of the kind of variables you have noted; yes, they have a time element to them. In order to get to depict them as a TIF graphic, you need to be able to process the netCDF file as was noted earlier by using some netCDF tool. Just about all statistical packages, including R, have tools to allow you to unpack the netCDF file and process it so that you can then later manipulate the data for selecting what you want to plot both spatially and temporally. Since you have familiarity with GIS software (and I have done this for people who have ArcGIS for example) then you should be able to export whatever you want in Excel or CSV format for input to the GIS software. In Matlab, which I use, below is how you would typically unpack the netCDF file, which generally have lat/lon spatial elements, a time element (in Julian format from a given tine - e.g., days since 01/01/01 - doing the ncdisp command below will tell you that) , and then the variable you are interested in; the first command will tell you what elements are in the .nc file and what the spatial gridding is

ncdisp ('September_2017_aviso_sla_data.nc')

lat = ncread ('September_2017_aviso_sla_data.nc', 'lat');

lon = ncread ('September_2017_aviso_sla_data.nc', 'lon');

time = ncread ('September_2017_aviso_sla_data.nc', 'time');

date = datevec(time); %converts the time from Julian to Gregorian

sla = ncread ('September_2017_aviso_sla_data.nc', 'sla');

At this point you can then manipulate the data to eventually create either a TIF file or Excel/CSV file for import into a GIS. If you don't have Matlab, then the following links can help you with getting this done as well; but frankly there is no direct way to convert from .nc to TIFF; there is a bit of processing required in between.

See:

http://nco.sourceforge.net/

http://research.jisao.washington.edu/data_sets/nco/

by GDAL: gdal_translate -of GTiff TEST.nc TEST.tiff

I stand corrected.

Thanks everyone for the suggestions. I'll be testing the methods in the next weeks and then I'll comeback with a feedback

This works well

library(ncdf4)

library(raster)

fname

You can convert nc files to Geotifs and geotifs to nc files using this on line site:

https://mygeodata.cloud/converter/netcdf-to-geotiff

Is there any way to read and process a large nc file (more than 20G)? I have tried IDL and MATLAB and both of them showed the file is too large and unreadable. Or any solutions to split it into small files? Thanks!

Jian Wang - You can give a try with Python. If your .nc file is gridded file, then you can split the file or separate variables or any other manipulations using CDO. It is a command-line tool very effective for the .nc files.

Hope this helps!

import os

import time

import glob

import arcpy

arcpy.env.overwriteOutput = True

#-------------------------------------------------------------------------------

inNetCDF = r"D:\GDB\ET\NTSG\Global_Monthly_ET_1982.nc"

outLoc = r"D:\GDB\ET\NTSG\MM"

#-------------------------------------------------------------------------------

allRasters = glob.glob(os.path.join(outLoc, '*.tif'))

allRasters.sort()

if len(allRasters) >= 1:

os.remove(allRasters[-1])

#-------------------------------------------------------------------------------

variable = "monthly_ET"

x_dimension = "lon"

y_dimension = "lat"

band_dimension = ""

dimension = "time"

valueSelectionMethod = "BY_VALUE"

nc_FP = arcpy.NetCDFFileProperties(inNetCDF)

top = nc_FP.getDimensionSize(dimension)

#-------------------------------------------------------------------------------

def str2(x):

x = str(x)

if len(x) != 2:

x = '0' + x

return x

#-------------------------------------------------------------------------------

for i in range(0, top):

dimension_values = nc_FP.getDimensionValue(dimension, i)

dimension_values = "time '{}'".format(dimension_values)

out_ras_lay = 'ET_1982_' + str(str2(i + 1))

out_ras = os.path.join(outLoc, "{}_.tif".format(out_ras_lay))

if not os.path.exists(out_ras):

print(out_ras)

arcpy.MakeNetCDFRasterLayer_md(inNetCDF, variable, x_dimension, y_dimension, out_ras_lay, band_dimension, dimension_values, valueSelectionMethod)

arcpy.CopyRaster_management(out_ras_lay, out_ras,"", "", "", "NONE", "NONE", "")

This can be done in R, quickly and efficiently,

First, use the ncdf4 package to pull out your variable and its corresponding lat/lon/time into a data frame, please follow this link

https://pravashtiwarigeop.wixsite.com/weathercode/post/netcdf-data-to-text-or-comma-seperated-variable-csv-in-r-easy-and-quick?fbclid=IwAR2tDp_qrF7Z3v4LSPbVgjm6i2ll6Z0H2gZ1s5IlRrFBeKX569RNTwVDWoY

for nc to csv conversion.

Once that is done, the conversion to .tiff can be carried out by the code below,

####read the csv#####

pts = read.table("file.csv",header = "TRUE)

library(sp)

library(rgdal)

coordinates(pts)=~lon+lat

r = raster(pts)

proj4string(pts) = CRS("+proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0")

pts = spTransform(pts,CRS("insert your proj4 string here"))

gridded(pts) = TRUE

r = raster(pts)

projection(r) = CRS("insert your proj4 string here")

plot(r)

writeRaster(r,"pts.tif")

This should do your work

Hope it helps !!

https://www.youtube.com/watch?v=2SdjVDldDxM

You may use

NCL: https://www.ncl.ucar.edu

or

GeoCAT: https://geocat.ucar.edu

You may use it

http://www.sdmtoolbox.org/

First

. You must install gdal library in your Linux system

Second

. You must select only one variable and one time step from your netcdf file

"cdo splitparam param input.nc output.nc"

"cdo splitday day input.nc output.nc"

Third

. Apply that command line ("gdal_translate -of GTiff file.nc test.tiff")

In Linux you can to try:

gdal_translate -of netCDF inputfile.tif outputfile.nc

(before you must install and load gdal)