Cutting Input files – in #10: INT2LM

in #10: INT2LM

<p> Yes, INT2LM can read global data cut out for a sub region. In older INT2LM version ( I am not sure until which version) it was an advantage to cut out the region before running the INT2LM since this would save time because the old versions always took the whole data into account for interpolation. In the versions we currently use the global data are cut in the INT2LM program first before the interpolation is done. <br/> If you have an account at <span class="caps"> DKRZ </span> you can do the cutting there, otherwise you need to transfer the data to your computer first. You may do the latter in a script month by month in order to keep the amount of disk space low at your institute. There is no automatic cutting procedure at <span class="caps"> DKRZ </span> presently. <br/> In oder to save space you can proceed as follows (assume for example your the original file is caf1992010100.nc): </p> <p> Use cdo or ncks to cut out the region you need. Actually cut out a bit larger region since the interpolation needs additional grid points at the boundaries. <br/> For a region with lower left (longitude1, latitude1) and upper right (longitude2, latitude2) perform either <br/> <pre> cdo sellonlatbox,longitude1,longitude2,latitude1,latitude2 caf1992010100.nc cas1992010100.nc ncks -h -d lon,longitude1,longitude2 -d lat,latitude1,latitude2 caf1992010100.nc cas1992010100.nc</pre> In case you use ncks: The coordinates should be <span class="caps"> REAL </span> values not <span class="caps"> INTEGER </span> , since <span class="caps"> INTEGER </span> values are interpreted as indices. </p> <p> You have to adopt the <span class="caps"> GRID </span> _IN part of the INT2LM namelist to match your region. To do this safely proceed as follows: <br/> Perform an <br/> <pre> ncdump -v lon,lat cas1992010100.nc</pre> From the information of the output change the following parameters in <span class="caps"> GRID </span> _IN accordingly: <br/> <code> ie_in_tot, je_in_tot, startlat_in_tot, startlon_in_tot, endlat_in_tot, endlon_in_tot </code> </p>

  @burkhardtrockel in #66dbfa0

<p> Yes, INT2LM can read global data cut out for a sub region. In older INT2LM version ( I am not sure until which version) it was an advantage to cut out the region before running the INT2LM since this would save time because the old versions always took the whole data into account for interpolation. In the versions we currently use the global data are cut in the INT2LM program first before the interpolation is done. <br/> If you have an account at <span class="caps"> DKRZ </span> you can do the cutting there, otherwise you need to transfer the data to your computer first. You may do the latter in a script month by month in order to keep the amount of disk space low at your institute. There is no automatic cutting procedure at <span class="caps"> DKRZ </span> presently. <br/> In oder to save space you can proceed as follows (assume for example your the original file is caf1992010100.nc): </p> <p> Use cdo or ncks to cut out the region you need. Actually cut out a bit larger region since the interpolation needs additional grid points at the boundaries. <br/> For a region with lower left (longitude1, latitude1) and upper right (longitude2, latitude2) perform either <br/> <pre> cdo sellonlatbox,longitude1,longitude2,latitude1,latitude2 caf1992010100.nc cas1992010100.nc ncks -h -d lon,longitude1,longitude2 -d lat,latitude1,latitude2 caf1992010100.nc cas1992010100.nc</pre> In case you use ncks: The coordinates should be <span class="caps"> REAL </span> values not <span class="caps"> INTEGER </span> , since <span class="caps"> INTEGER </span> values are interpreted as indices. </p> <p> You have to adopt the <span class="caps"> GRID </span> _IN part of the INT2LM namelist to match your region. To do this safely proceed as follows: <br/> Perform an <br/> <pre> ncdump -v lon,lat cas1992010100.nc</pre> From the information of the output change the following parameters in <span class="caps"> GRID </span> _IN accordingly: <br/> <code> ie_in_tot, je_in_tot, startlat_in_tot, startlon_in_tot, endlat_in_tot, endlon_in_tot </code> </p>
BR replied

Yes, INT2LM can read global data cut out for a sub region. In older INT2LM version ( I am not sure until which version) it was an advantage to cut out the region before running the INT2LM since this would save time because the old versions always took the whole data into account for interpolation. In the versions we currently use the global data are cut in the INT2LM program first before the interpolation is done.
If you have an account at DKRZ you can do the cutting there, otherwise you need to transfer the data to your computer first. You may do the latter in a script month by month in order to keep the amount of disk space low at your institute. There is no automatic cutting procedure at DKRZ presently.
In oder to save space you can proceed as follows (assume for example your the original file is caf1992010100.nc):

Use cdo or ncks to cut out the region you need. Actually cut out a bit larger region since the interpolation needs additional grid points at the boundaries.
For a region with lower left (longitude1, latitude1) and upper right (longitude2, latitude2) perform either 

cdo sellonlatbox,longitude1,longitude2,latitude1,latitude2 caf1992010100.nc cas1992010100.nc
ncks -h -d lon,longitude1,longitude2 -d lat,latitude1,latitude2 caf1992010100.nc cas1992010100.nc
In case you use ncks: The coordinates should be REAL values not INTEGER , since INTEGER values are interpreted as indices.

You have to adopt the GRID _IN part of the INT2LM namelist to match your region. To do this safely proceed as follows:
Perform an 

ncdump -v lon,lat cas1992010100.nc
From the information of the output change the following parameters in GRID _IN accordingly:
ie_in_tot, je_in_tot, startlat_in_tot, startlon_in_tot, endlat_in_tot, endlon_in_tot

View in channel