Yes, you can use the function "filwgts_lancos" to calculate weights for a
Lanczos filter, then the function "wgt_runave" to calculate a weighted
running average using the filter. For information on using these
functions, see the web pages:
Typically, what I do is use an 8-day high pass filter on daily mean
variables to retain variability on approximately 2-8 day time scales for
my storm track analysis. So, first I create the weights for a 15-point,
8-day high pass filter with NCL code such as the following:
befaft = 7 ; number of points to use before and after current point
numwts = 2*befaft + 1 ; total number of weights
ihp = 1 ; use high pass filter
fca = 0.125 ; cutoff at 8 day period
fcb = -999. ; don't use second cutoff
nsigma = 1. ; "nsigma = 1. is common"
wts = filwgts_lancos(numwts,ihp,fca,fcb,nsigma)
Then, if I have daily 300 mb height data in the variable Z300 with
dimensions Z300(time,lat,lon), I filter the data using code such as:
filt = wgt_runave_Wrap(Z300(lat|:,lon|:,time|0:2*befaft+daysinmonth,wts,0)
Z300_filt = filt(:,:,befaft:befaft+daysinmonth-1)
noting that I need befaft=7 days before and after the days for which I
want filtered data.
Let me know if you have more questions.
On Thu, 2 Dec 2004, Shih-Yu Lee wrote:
> Is NCL has any build in function that could do bandpass filter? Is
> anybody can tell me how to do storm track analysis? The only thing I
> have is the geopotential height field in netcdf file. How can I do storm
> track analysis?
ncl-talk mailing list
ncl-talk AT ucar.edu
This archive was generated by hypermail 2b29 : Tue Dec 07 2004 - 16:33:07 MST