[ncl-talk] principal oscillation pattern

[Maria Gehne - NOAA Affiliate]

Ok, I remembered something about the correlation matrices. In version 6.1.2
there was a bug in
https://www.ncl.ucar.edu/Document/Functions/Built-in/covcorm.shtml
where the options behaved the opposite of what they were supposed to do.
Maybe that's the issue here.
I would recommend downloading the POP driver script
http://ncl.ucar.edu/Applications/Scripts/PrnOscPat_driver.ncl
save it under a different name and load that script at the beginning of
your code. Instead of calling the built-in version, call the new version in
your code.

Before you do that you can change the options for covcorm in line 25 in the
new POP driver script. I would also include a few printVarSummarys
throughout the code so that you can see if your array is suddenly 1D
instead of 2D.

Hope this helps,
Maria

On Thu, Jan 8, 2015 at 3:33 PM, Maria Gehne - NOAA Affiliate <
maria.gehne at noaa.gov> wrote:

> Hi Lena,
> I can look into it tonight or tomorrow morning. I'm traveling right now so
> it's hard to look at your code and the pop routines. I wrote most of the
> original code so hopefully I can help you.
>
> Maria
>
>
> On Thursday, January 8, 2015, Dennis Shea <shea at ucar.edu> wrote:
>
>> Technically,the 'Principal Oscillation Pattern' is not a supported
>> function. However, it will be looked at in th near future: today or
>> tomorrow.
>>
>> On Thu, Jan 8, 2015 at 11:22 AM, Mary Haley <haley at ucar.edu> wrote:
>>
>>> Lena,
>>>
>>> A friendly tip: if you reply to an email thread, it's helpful to keep
>>> all the original emails in the body of the message. This particular email
>>> didn't include the original email with the error message, so I had to go
>>> back a few emails to find it.
>>>
>>> The error is referring to an "inverse_matrix" function, but your script
>>> is not calling this function directly, so it's difficult for us to debug
>>> your script without knowing what other function is referencing
>>> inverse_matrix.
>>>
>>> Usually you can debug these things on your own by using
>>> "printVarSummary", which will tell you what the dimensionality and size of
>>> your data is.
>>>
>>> Find the location in your script where you think inverse_matrix is being
>>> called by another function, and then insert some "printVarSummary" calls to
>>> look at the data before you call the offending function. This will
>>> hopefully shed some light on why your data is not the correct
>>> dimensionality.
>>>
>>> --Mary
>>>
>>>
>>>
>>>
>>> On Thu, Jan 8, 2015 at 6:43 AM, Lena Frey <Frey_L at web.de> wrote:
>>>
>>>> Hi again!
>>>>
>>>> Thanks to Karin. The problem is, that I don't know why I get this error
>>>> message.
>>>> I have used the ncl script from the ncl website to calculate the POPs
>>>> for the mean sea level pressure.
>>>> I am using NCEP reanalysis data. The mean sea level pressure has the
>>>> dimensions time, lat, lon.
>>>> So, following the ncl script it should be 2D before calculating the
>>>> inverse matrix.
>>>>
>>>> You can find attached my ncl script.
>>>>
>>>> Thanks,
>>>>
>>>> Lena
>>>>
>>>> _______________________________________________
>>>> 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
>>>
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20150108/2572156b/attachment.html