Post-process SNODAS data

I will show how to post-process SNODAS data using a suite of tools in the command line. This tutorial has only been tested on Linux/Ubuntu system, but the method should be the same for other OS.

Some notes about the SNODAS snow data:

• The User guide provides the most complete information about SNODAS.
• the masked snow data files for the CONUS since Oct 2003; the unmasked data is for the

Software requirement

nco, Docker

Download data

See https://nsidc.org/support/64231694-FTP-Client-Data-Access for more details.

I used wget to download all the snow data:

wget --ftp-user=anonymous -r -nd ftp://sidads.colorado.edu/DATASETS/NOAA/G02158/masked

Post-process

Untar and unzip

# untar
ls *.tar | xargs -i tar xf {}

# unzip .gz files
gunzip *.gz  # note, this will replace original files!

You will have both *.dat and *.txt file with the same file name.

Each .dat file is differentiated by the product code. See SNODAS user guide for more details.

• us_ssmv11034tS*.dat is for SWE

Subset data using pyrs

See https://nsidc.org/support/how/subsetting-and-reformatting-nsidc-pyrs-docker-container for more info.

• prepare data. Move all us_ssmv11034tS__T0001TTNATS*.dat into a single folder.

• Download docker (need root permission). Follow the instructions for installing Linux-ubuntu version.

• Run docker. This will mount input datasets as well as output directory, and run pyrs interactively.

• docker run -it --rm -v /mnt/4tbd/pin/SNODAS/extracted-files/us_swe/extracted-files/dat/:/DATASETS -v /mnt/4tbd/pin/SNODAS/extracted-files/us_swe/extracted-files/netcdf/:/output  nsidc/pyrs_container:v1.1.0
  

  • Run PyRS

    # enter command for interactive operation
    $ pyrs
    # choose data source
    Source type (one of SNODAS, NSIDC-0051, NSIDC-0081):SNODAS
    # choose format
    Output format (one of geotiff, netcdf): netcdf
    # then follow the prompts to enter input/output directory
    Input directory [/]:/DATASETS
    Output directory [/]:/output
    # wheter to subset the data
    Would you like to subset the data? [y/N]:y
    # enter the box region in decimal
    westernmost bound (decimal degrees from -125.0 to -67.0) [-125.0]:-108
    northernmost bound (decimal degrees from 25.0 to 53.0) [53.0]:40
    easternmost bound (decimal degrees from -108.0 to -67.0) [-67.0]:-106
    southernmost bound (decimal degrees from 25.0 to 40.0) [25.0]:38
    # wait for it to process
    # exit when it is done
    exit
    

combine netcdf files

• download NCO tookit For Ubuntu/Linux:

  aptitude install nco
  

• Combine all daily netcdf files using ncrcat command:

ls | grep \.nc | ncrcat -o SWE_2014-east-taylor.nc

• remove concat history from the attributes field using ncatted command. This can help to reduce the netcdf filesize.

ncatted -a nco_input_file_list,global,d,, SWE_2014-east-taylor.nc

​ For details on how to use ncatted, see http://nco.sourceforge.net/nco.html#ncatted

​ - view the contect of netcdf using ncdump -hs SWE_2014.nc

Subset region

Use ncea to subset regions. See documentation for more details.

# format
ncea -d lat,minimum_lat,maximum_lat -d lon,minimum_lon,maximum_lon in.nc out.nc
# for example
ncea -d lat,38.0,40.0 -d lon,-108.0,-106.0 SNODAS_SWE-US-2003-2021.nc SNODAS_SWE_EastTaylor-2003-2021.nc