Re: create longitude and latitude in netcdf

From: Dennis Shea <shea_at_nyahnyahspammersnyahnyah>
Date: Thu Jun 27 2013 - 09:08:59 MDT

[1] The 'new' file has variables named x, y, lat and lon.
     Each of these variables has a dimension named 'land'
     *None* of these is a "coordinate variable" (CV) because they
     are not monotonic.

[2] To create a grid which is filled with the 'tair' at
     each lat/lon point requires that you create a grid template
     and fill the template with each point.

See attached.

The created file has two CVs: LAT and LON

netcdf tair_2010_model_1time_newlonlat.grid {
dimensions:
        LAT = 180 ;
        LON = 360 ;
variables:
        float tair(LAT, LON) ;
                tair:long_name = "temperature" ;
                tair:units = "K" ;
                tair:_FillValue = -1.e+34f ;
        float LAT(LAT) ;
                LAT:units = "degrees_north" ;
        float LON(LON) ;
                LON:units = "degrees_east" ;

Good luck

On 6/26/13 5:12 PM, Jiaxin Zhang wrote:
> Thank you so much, Dennis!
>
>
>
> I have modified the ncl code (deleting .0 gives integer values in the y
> calculation) and now the lon computation is correct (attached file).
>
>
>
> But I still have one issue. I need to add lat and lon as coordinate
> variables for air temperature. According to the second condition that you
> mentioned, lat and lon should be not only variables but also dimensions,
> right? The new netcdf has only 1 dimension (land), and I guess it would be
> necessary to add lat and lon as dimensions. Then, probably all 3
> requirements would be satisfied for both lat and lon, and it would be
> possible to add them as coordinate variables for air temperature.
>
>
>
> I was trying to add lat and lon as dimensions in the last part of the first
> code version that I sent (the strange part of the code that you mentioned),
> but it did not work (problems with dimensions). I guess there is a way to
> do it, but I did not find it in the NCL documents that I checked. How is it
> possible to add lat and lon as dimensions in the file? How is it possible
> to add lat and lon as coordinate variables for air temperature? What has to
> be modified in the attached ncl code to do this part successfully?
>
>
>
> Thank you!
>
>
>
> Jiaxin Zhang
>
>
> 2013/6/26 Dennis Shea <shea@ucar.edu>
>
>> The script you attached is rather confusing.
>> Basically, the file contains land only grid pts
>> in a one dimensional array. The lat and lon values
>> can only be reconstructed for the land points.
>>
>> The file has a varaible
>> land(land)
>>
>> This is known as a 'coordinate variable' (CV) in netCDF-speak
>>
>> http://www.unidata.ucar.edu/**software/netcdf/docs/**
>> coordinate_variables.html<http://www.unidata.ucar.edu/software/netcdf/docs/coordinate_variables.html>
>>
>> Requirements for a CV
>> (1) one dimensional
>> (2) dimension name ('land') is the same as the variable name
>> [land(land)]
>> (3) numeric, monotonically increasing or decreasing values
>>
>> The land(land) satisfies the first two conditions but fails [3].
>>
>> CVs have a special status in NCL. Since this is not a CV the
>> attached script addresses the issue.
>>
>> ==
>> No idea why you did
>>
>> temp!0 = "newland"
>> temp!1 = "newlat"
>> temp!2 = "newlon"
>>
>> temp&newland = land
>> temp&newlat = lat
>> temp&newlon = lon
>> ===
>>
>> NOTE: the lat computation looks good. The lon computation is
>> not correct. It is your responsibility to fix that.
>> Always examine the output from printVarSummary(...)
>>
>>
>>
>>
>> On 6/25/13 5:26 PM, Jiaxin Zhang wrote:
>>
>>> Hi,
>>>
>>>
>>>
>>> I am a new NCL user. After having a look at the NCL Mini-Language Manual
>>> and other resources in the NCL website I still could not find a solution
>>> for this problem.
>>>
>>>
>>>
>>> I have a netcdf file (attached file "tair_2010_model_1time.nc") with air
>>> temperature values and the only coordinates in the file are land numbers
>>> (calculated from longitude and latitude values). However, I need to add
>>> the
>>> traditional coordinates with longitude and latitude associated with air
>>> temperature values in the netcdf file.
>>>
>>>
>>>
>>> To calculate back longitude and latitude values from land values, there
>>> are
>>> the following equations:
>>>
>>>
>>>
>>> y = INT( (land 1) / 360) ! lat index from 0 to 179
>>>
>>> x = land 360 * y ! lon index from 1 to 360
>>>
>>> lat = 89.5 y
>>>
>>> lon = -179.5 + x 1
>>>
>>>
>>>
>>> I tried to do these calculations and create the variables y, x, lat and
>>> lon
>>> in a new netcdf file and apparently this part works (can anyone please let
>>> me know if this part is correct in the attached ncl code "newlonlat.ncl"?
>>> It is the first time I am creating new variables in a netcdf file and I am
>>> not completely sure that this is the best way to do it. I have used the
>>> function "toint" to convert y to integer values). But one of the main
>>> problems seems to be related with the number of dimensions. The original
>>> netcdf has only 1 dimension (land), whereas in the new netcdf, I would
>>> need
>>> to have more dimensions (at least longitude and latitude), and it seems
>>> that this gives an error message. Also, longitude and latitude should be
>>> coordinate variables for air temperature in the new netcdf file.
>>>
>>>
>>>
>>> Could anyne please help me to find where the error lies and modify the
>>> attached script to create the correct netcdf file, including new values
>>> for
>>> longitude and latitude as coordinate variables for air temperature?
>>>
>>>
>>>
>>> Any help will be very appreciated!
>>>
>>>
>>>
>>> Thank you!
>>>
>>>
>>>
>>> Jiaxin Zhang
>>>
>>>
>>>
>>> ______________________________**_________________
>>> ncl-talk mailing list
>>> List instructions, subscriber options, unsubscribe:
>>> http://mailman.ucar.edu/**mailman/listinfo/ncl-talk<http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
>>>
>>>
>
>
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

Received on Thu Jun 27 09:09:13 2013

This archive was generated by hypermail 2.1.8 : Mon Jul 01 2013 - 12:35:42 MDT