Re: average & std. deviation for non-contiguous time slices

On 03/12/2014 11:02 AM, Ray Sambrotto wrote:
>
> I am trying to implement a statistical comparison similar to the one in
> the climo1.ncl example script. The problem is that I am working with
> satellite imagery that is irregularly spaced in time and not monthly
> climatology so I don’t think I can use clmMonLLT & stdMonLLT. Also, the
> dim_sum_n function requires all of the data to be in the same file,
> which these are not. I suppose I could construct such a file from the
> separate data files.
>
> My question is: Is there a better method for finding the average and
> standard deviation of a non-contiguous group of time slices?
>
> Thanks, Ray Sambrotto
>
> -----------
> Lamont-Doherty Earth Observatory
> 61 Rt. 9W / Palisades, NY 10964 USA
>
Hi Ray

I am not familiar to satellite data format,
but I guess you could use a variable on a regular grid with
missing values on the data gaps.

clmMonLLT uses dim_avg, and stdMonLLT uses dim_stddev,
both of which claim to ignore missing values in their calculations.

Say, using "new", you could create a data(lat,lon,time)
variable with a _FillValue of your choice (or perhaps better,
the same _FillValue that is used on your files).
"data" should be large enough to hold all actual data
(i.e. have maximum spatial and time coverage).
You probably need to create "named" dimensions for it too.

Then fill in the "data" variable using the various files,
but honoring the (spatial and) time gaps the data may have.
This may depend on how the data is organized on the files
(one or more time records per file, full or partial spatial coverage
per file, etc).

Finally apply the clmMonLLT and stdMonLLT to
the "data" variable.

I hope this helps,
Gus Correa

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Mar 12 11:04:18 2014