Re: spectral Analysis

Hi Paul,

An example on how to get the frequency limits depending on the time step
and frequency cut-off is here:
http://ncl.ucar.edu/Document/Functions/Built-in/filwgts_lanczos.shtml
and here
http://www.ncl.ucar.edu/Applications/filter.shtml

In your case I believe you need to set fca = 4/60 to get the correct
frequency cut-off.

Hope this helps,
Maria


On Sat, Jul 19, 2014 at 6:14 AM, Steeven Paul Yerraguntla <
steevenpaul@gmail.com> wrote:

> Thanks to Maria for valuable information. I wanted to filter my data above
> 1 hour (60 min) and the data is in the 4 minute time step. For that i tried
> to 60 min low pass the data using Lanczos funtion as follows.
>
>
> *****************************************************************************************
> ihp = 0 ; Low pass
> sigma = 1.0 ; Lanczos sigma
>
> nWgt = 1201
> * fca = 1./(4*60) * ; start freq
> fcb = -999 ; last freq
> wgt = filwgts_lanczos (nWgt, ihp, fca, fcb, sigma)
> xBPF = wgt_runave_Wrap ( x, wgt, 0 )
>
> ***************************************************************************************
> Is fca assignment meets my requirement otherwise please let me know the
> right calculation.
>
> Thanks in advance
>
>
> On Fri, Jul 18, 2014 at 10:55 PM, Maria Gehne - NOAA Affiliate <
> maria.gehne@noaa.gov> wrote:
>
>> Hi Paul,
>>
>> when you do a spectral analysis the lowest frequency you can resolve is
>> 1/N if N is the length of your time series. The highest frequency is
>> 0.5/10mins in your case, so half a cycle in 10 minutes. Or in general half
>> a cycle in the time interval. The frequency units will be 1cycle/time unit
>> so in your case cycles/10min. To get to cycles/min divide the frequency
>> vector by 10.
>>
>> It helps if you know certain signals in your data and what frequency they
>> should be at. That's a good sanity check when computing spectra to make
>> sure the frequency units are correct.
>>
>> Hope this helps,
>> Maria
>>
>>
>> On Fri, Jul 18, 2014 at 3:58 AM, Steeven Paul Yerraguntla <
>> steevenpaul@gmail.com> wrote:
>>
>>> Dear NCL users,
>>> I have plotted spectral variance to my data using Data
>>> Analysis examples in the NCL website. The data i have used is of 10 min
>>> interval and my doubt is that whether spectral frequency would be
>>> cycles/min or cycles/10min. If it is not cycles/min then how can i get
>>> the spectral Frequency in cycles/min. Hope i will get the answer soon.
>>> Thanks in advance.
>>>
>>> regards,
>>> Paul.
>>>
>>> _______________________________________________
>>> ncl-talk mailing list
>>> List instructions, subscriber options, unsubscribe:
>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>>
>>>
>>
>

Received on Mon Jul 21 03:43:42 2014