[ncl-talk] serious bugs of fourier_info in NCL v.6.2.0

Mary Haley haley at ucar.edu
Wed Mar 4 15:03:09 MST 2015


Thomas,

I did discover that a change was made to an internal Fortran routine before
V6.2.0 was release, but I can't verify if this is the source of the
potential problem.

It would be helpful if you could provide the data that you are passing to
fourier_info, and the values of the other two parameters, so we can verify
the bug here.  The fourier_info tests I have are not showing any issues in
V6.2.0.

Thanks,

--Mary


On Mon, Mar 2, 2015 at 11:19 PM, Thomas Tobian <thomastobian at gmail.com>
wrote:

> If you want to check this,  just simply test it with any data you have. I
> tested also with NCL v.6.0.0 it works fine as in the v.6.1.2. This is only
> incorrect with 6.2.0.
>
> No, there is no issue with scaling factor. Yes this is the absolute
> values, not responses or anomalies.
>
> Thomas
>
>
> On Tuesday, March 3, 2015, Maria Gehne - NOAA Affiliate <
> maria.gehne at noaa.gov> wrote:
>
>> From your figures it also looks like the patterns could be the same and
>> that just the scaling is different. Did you look at the actual values?
>>
>> Maria
>>
>> On Mon, Mar 2, 2015 at 5:36 PM, Dennis Shea <shea at ucar.edu> wrote:
>>
>>> Just sending these figures would indicate an issue. However, it does not
>>> help in finding the source.
>>> Checking the history of functions ...
>>>    http://www.ncl.ucar.edu/prev_releases.shtml
>>>
>>> This does not indicate that any explicit change has occurred in
>>> 'fourier_info' since its introduction.
>>>
>>> Are the number of  values odd or even ...
>>> Really, nothing can be done without more information.
>>>
>>> On Mon, Mar 2, 2015 at 3:27 PM, Thomas Tobian <thomastobian at gmail.com>
>>> wrote:
>>>
>>>> Dear NCL developer,
>>>>
>>>> I found serious bugs for the fourier_info function in NCL v.6.2.0,
>>>> Attached, please find an example to illustrate it:
>>>> The first two rows are the amplitude of ozone wave1 based on NCL. v.
>>>> 6.1.2
>>>> and the 3rd-4th rows are computed based on NCL v.6.2.0? The results
>>>> from v.6.2 seems to be not realistic and incorrect.. Could you please find
>>>> the source of the errors? This case also applies for the phase and
>>>> explained variance results.
>>>>
>>>> Thanks
>>>> Thomas Tobian
>>>>
>>>> _______________________________________________
>>>> 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
>>>
>>>
>>
> _______________________________________________
> 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/20150304/d3b60c14/attachment.html 


More information about the ncl-talk mailing list