Re: [ferret_users] reading ascii data and regridding

["William S. Kessler" <William.S.Kessler@xxxxxxxx>]

Thanks again to Jaison for his typically thorough discussion.

This might be one of those cases where it is better to make an  
intermediate file for checking before proceeding to a final  
interpolation. I would write out the (combined) profiles made in  
Jaison's step 2 as a grid with dimensions (z,t). There are several  
reasons for this: First, to plot the z-interpolated profiles against  
the original data, to be sure that the vertical resolution is  
sufficient to capture the observations (especially at the top and  
bottom ends, where some profiles may be missing data); Second,  
because profile data sets like this often contain spikes or other  
data errors, which should be caught before they contaminate a gridded  
product; Third, because the horizontal gridding requires choices that  
will produce quite different results, some experimentation will be  
required in Jaison's step 3.

Use the method described in Chapter 2, Section 10 of the manual,  
Example 3 (appending multiple timesteps), to create an intermediate  
netcdf file with the data on a (regular-z,index-t) grid. That is,  
each time you read and vertically-interpolate a profile, append it to  
the netcdf file using a "timestep" as the profile number. Include the  
lat/lon information on the same index-t grid. With that file in hand  
(and checked against the original), you can easily experiment with  
the horizontal gridding algorithms.

In general, it is better not to proceed blindly through several steps  
of processing irregular data to a final result, but to stop and  
reflect at each major step. After producing the intermediate (z,t)  
file, make many test plots, check variances and any other statistics  
you can, to be sure that you are not propagating errors or faulty  
assumptions into your gridded product.

Billy K

On Jun 12, 2007, at 7:27 AM, Jaison Kurian wrote:

> Hello Sriram,
>                Your mail have 3 questions. I will try to answer them
> separately.
>
> Assuming :
>      -  all variables are in "numeric format", including longitude and
>            latitude (ie. no "N" or "E" associated with the values)
>      -  no comments anywhere in the datafile,
>      -  data is for 1 particular float, and for a particular profile
>
> 1. Reading ASCII data
> ---------------------
>
>    define axis/Z=0:100:1/units=meters/DEPTH zfile
>    define grid/z=zfile gfile
>    FILE/grid=gfile/var="fid, yr, mon, date, lon, lat, press, te,  
> sa" ascii.txt
>
> 2. Interpolating to standard depth levels
> -----------------------------------------
>
>    ! first regrid the temp. data onto pressure axis (units dbar/ 
> meters)
>
>    define axis/z/units=meters/DEPTH/from_data zpress=press[d=1]
>    let te_pr = te[d=1,gz=zpress@ASN]
>
>    ! now, on to the destination standard depth axis
>
>    let zstdr = ZSEQUENCE 
> ({0,10,20,30,40,50,75,100,150,200,300,400,500,750,1000})
>    define axis/z/units=meters/DEPTH/from_data zstnd=zstdr
>    let temp  = te_pr[gz=zstnd]   ! the default linear interpolation  
> is assumed
>
> 3. Profiles to X-Y Gridded data
> --------------------------------
>
>    To remap the profile data (on standard depth or pressure axis)  
> on to a
>    uniformly spaced XY grid, you have to use external functions like
>    SCAT2GRIDGAUSS_XY or SCAT2GRIDLAPLACE_XY. For this purpose, you  
> need the
>    lat and lon information for each profile.
>
> These are the general steps you have to go. Steps 1 and 2 will  
> depend on the
> "exact" format of the ascii data when you have multiple profiles.  
> For example,
> if the pressure levels are different for different profiles, then  
> you have to
> repeat the definition of zpress for all the profiles.
>
> You can incorporate the time axis information during any of the  
> above steps
> or at the final stage.
>
> Relevant sections in user manual are :
>
>   Ch2 Sec5.1.    Reading ASCII files
>   Ch2 Sec5.2.    Reading "DELIMITED" data files
>   Ch2 Sec6.      TRICKS TO READING BINARY AND ASCII FILES
>   Ch3 Sec2.3.37  SCAT2GRIDGAUSS_XY
>   Ch3 Sec2.3.40  SCAT2GRIDLAPLACE_XY
>
> In Ferret Mail archives, you can find many threads discussing these  
> topics.
> However, the "Search the Archives" tool at Ferret website is not  
> working I
> guess. But you can try searching in google.
>
> Hope this helps,
>
> Regards,
>
> Jaison
>
>
> On Mon, 11 Jun 2007, Prakya Shree Ram wrote:
>
> > Dear Friends,
> >
> > I am trying to read ascii data which contains the
> > vertical profiles of temperature and salinity. My data
> > file contains 9 columns (float ID, year, month, date,
> > longitude, latitude, pressute, temperature and
> > salinity). The observations are at random locations, I
> > want to do interpolation of temperature and salinity
> > for standard depths and then into 0.25 X 0.25
> > horizontal. Can anybody suggest me the best possible
> > way to do this?
> >
> > Any help in this regard is highly appreciated.
> >
> > Thanks in advance and regards,
> > Sriram
> >
> >
> >
> > _____________________________________________________________________ 
> > _______________
> > Sucker-punch spam with award-winning protection.
> > Try the free Yahoo! Mail Beta.
> > http://advision.webevents.yahoo.com/mailbeta/features_spam.html
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.