>> I found that the spectrum plot will be changed
>> greatly if I standardize the time series before
>> using "specx_anal".
>>
>> I am a little puzzled on it. Is it necessary to
>> do a standardization at first when we analyze the spectrum ?
>>
>> ( I analyze the 102-yrs precipitation datas , all data ,
>> script and plots have been uploaded to ftp://cgd.ucar.edu/incoming/ )
>-------------
>
>Standardization normalizes the data [variance=1].
>
>The point of spectral analysis is to look at which frequency
>bands have the most variance and are they significant
>either physically or statistically.
>
>Don't normalize before doing spectral analysis.
=====
All ...
[1] I apologize for the above response.
I did not look at the script before I replied!
More important, I did not think before I replied :-(
[2] The spectral pattern should be the same
whether the data are normalized or not.
Of course, the magnitudes will be different.
[3] Before responding this time, I decided to
look at the script that was sent. :-)
The original code had:
prcp = asciiread("foo.dat",(/102,8/),"float")
prcp = dim_standardize(prcp,1)
As noted in the "dim_standardize" documentation the
*rightmost dimension* is being operated upon.
(like all the dim_* functions)
Based upon the script the left dimension represents "time"
and the right "stations"
Does anyone see the problem?!?!
Answer: the dimensions must be reordered to get the standardized
time series at each station. The following is suggested
PRCP = asciiread("foo.dat",(/102,8/),"float")
PRCP!0 = "time"
PRCP!1 = "station"
prcp = PRCP(station|:,time|:) ; make time the rightmost dimension
prcp = dim_standardize(prcp,1) ; for each station
; standardize over time
D
D
_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Thu Feb 08 2007 - 11:50:13 MST
This archive was generated by hypermail 2.2.0 : Mon Feb 12 2007 - 09:02:03 MST