[Thread Prev][Thread Next][Index]

Re: [ferret_users] filters

Hi Kalyani,
The weights are computed as follows


      nn = (N-1)/2
      weight(0)  = 2.*(f2-f1)
      do k = 1,nn
         pik   = pi*k
         aa    = pik/nn
         sigma = sin(aa)/aa
         aa    = 2.*pik*f1
         bb    = 2.*pik*f2
         weight( k) = ( sin(bb)/pik - sin(aa)/pik )*sigma
      end do

And then they are applied as follows (arg_1 is the data, and there are other loops over the x, y, z dimensions of arg_1, which accounts for the i, j, k indices)
        DO 100 l=res_lo_ss(T_AXIS)+nn, res_hi_ss(T_AXIS)-nn

           do m = 1,nn
                   result(i,j,k,l) = result(i,j,k,l) + weight(m)*
     .                   ( arg_1(i,j,k,l-m) + arg_1(i,j,k,l+m) )
           end do
100     end do

C Kalyani Devasena wrote:
Dear ferret users,

I am using ferret function LANCZOS (filter), band-pass filter for analysising a time series data (trition buoy data) for identifying 6-30 days and 30-90 day intraseasonal oscillation. The data is daily and I have 360 days of the data

I am not able to understand how ferret is selecting the weights.

May I request for some needy help!!!

How to select correct weights in using the filter.

Ex :!for 30-90days
plot lanczos(az,0.0111,0.033,91) or plot/over lanczos(az,0.0111,0.033,61) shows lot of difference.

I have some confusion  in understanding the funtion application.

with regards,

[Thread Prev][Thread Next][Index]

Contact Us
Dept of Commerce / NOAA / OAR / PMEL / TMAP

Privacy Policy | Disclaimer | Accessibility Statement