The file topodata_0.9x1.25_USGS_070110_stream_c151201.nc is identical to
lnd/clm2/griddata/topodata_0.9x1.25_USGS_070110.nc, but with TOPO having a time
dimension, and added mask and area fields, so that it can be read as a stream
(with domain information in the same file). It was modified as follows:

----

In python:

from netCDF4 import Dataset
topodata = Dataset('topodata_0.9x1.25_USGS_070110_stream_c151201.nc', 'a')

topodata.createDimension('time', size=1)
topodata.createVariable('time', 'f8', ('time'))
topodata.variables['time'].long_name = "time"
topodata.variables['time'].units = 'days since 0001-01-01 00:00:00'
topodata.variables['time'].calendar = 'noleap'
topodata.variables['time'][:] = 0.

topodata.renameVariable('TOPO', 'TOPO_old')
topodata.createVariable('TOPO', 'f8', ('time','lsmlat','lsmlon'))
topodata.variables['TOPO'][:] = topodata.variables['TOPO_old'][:]
topodata.variables['TOPO'].long_name = "topography height"
topodata.variables['TOPO'].units = 'm'

topodata.createVariable('mask', 'i4', ('lsmlat','lsmlon'))
topodata.variables['mask'][:] = 1
topodata.variables['mask'].long_name = "mask of valid data (0 = invalid, 1 = valid)"
topodata.variables['mask'].units = "unitless"

domain = Dataset('/glade/p/cesmdata/cseg/inputdata/share/domains/domain.lnd.fv0.9x1.25_gx1v6.090309.nc')
topodata.createVariable('area', 'f8', ('lsmlat','lsmlon'))
topodata.variables['area'][:] = domain.variables['area'][:]
topodata.variables['area'].long_name = "grid cell area"
topodata.variables['area'].units = 'radian2'

topodata.renameDimension('lsmlat','lat')
topodata.renameDimension('lsmlon','lon')

topodata.close()

----

Removing old TOPO variable using nco, because there is no way to delete a
variable using python's netCDF4:

ncks -x -v TOPO_old topodata_0.9x1.25_USGS_070110_stream_c151201.nc topodata_new.nc
mv topodata_new.nc topodata_0.9x1.25_USGS_070110_stream_c151201.nc