Re: [python-users] How to make scale_factor work with python-netcdf4 + np.ushort

• To: python-users@xxxxxxxxxxxxxxxx
• Subject: Re: [python-users] How to make scale_factor work with python-netcdf4 + np.ushort
• From: daryl herzmann <akrherz@xxxxxxxxxxx>
• Date: Mon, 26 Jun 2017 10:13:57 -0500 (CDT)
• Authentication-results: unidata.ucar.edu; dkim=none (message not signed) header.d=none;unidata.ucar.edu; dmarc=none action=none header.from=iastate.edu;
• Spamdiagnosticmetadata: NSPM
• Spamdiagnosticoutput: 1:99

Greetings,

https://github.com/Unidata/netcdf4-python/issues/682

daryl

On Sat, 17 Jun 2017, daryl herzmann wrote:

Howdy,

Would some kind python user hit me with the clue-by-four here as to how to make scale_factor work with python-netcdf4 and the ushort type? I am using current conda-forge on python2.7. Here's my minimal example:

"""scale_factor.py"""
from __future__ import print_function
import netCDF4
import numpy as np


def create_file():
    """Create"""
    nc = netCDF4.Dataset('test.nc', 'w')
    nc.createDimension('x', 10)
    ncvar = nc.createVariable('ncvar', np.ushort, ('x', ))
    ncvar.scale_factor = 100.
    ncvar.add_offset = 0.
    nc.close()


def write_data():
    """Write"""
    nc = netCDF4.Dataset('test.nc', 'a')
    nc.set_auto_maskandscale(True)
    nc.variables['ncvar'][5] = 1.5
    nc.close()

    nc = netCDF4.Dataset('test.nc', 'a')
    nc.set_auto_maskandscale(False)
    nc.variables['ncvar'][6] = 150.5
    nc.close()


def read_file():
    """Read"""
    nc = netCDF4.Dataset('test.nc', 'r')
    nc.set_auto_maskandscale(True)
    print("read[5] resulted in %s" % (nc.variables['ncvar'][5],))
    print("read[6] resulted in %s" % (nc.variables['ncvar'][6],))
    nc.close()

    nc = netCDF4.Dataset('test.nc', 'r')
    nc.set_auto_maskandscale(False)
    print("readv2[5] resulted in %s" % (nc.variables['ncvar'][5],))
    print("readv2[6] resulted in %s" % (nc.variables['ncvar'][6],))
    nc.close()


def main():
    """Go Main"""
    create_file()
    write_data()
    read_file()


if __name__ == '__main__':
    main()

This results in

$ python scale_factor.py
read[5] resulted in 0.0
read[6] resulted in 15000.0
readv2[5] resulted in 0
readv2[6] resulted in 150

if I switch to np.float32 as the netcdf variable type, I get proper results

$ python scale_factor.py
read[5] resulted in 1.49999996647
read[6] resulted in 15050.0
readv2[5] resulted in 0.015
readv2[6] resulted in 150.5


thanks!
daryl

--
/**
 * daryl herzmann
 * Systems Analyst III -- Iowa Environmental Mesonet
 * https://mesonet.agron.iastate.edu
 */

• References:
  • [python-users] How to make scale_factor work with python-netcdf4 + np.ushort
    • From: daryl herzmann