As for ttest, when two original samples are assumed
to have the same population variance, the parameter
iflag is set to True in ttest function in NCL, and
subroutine ttest.for is used in Numeric Recipi,
whereas the two original samples are assumed to have
the different population variance, the parameter iflag
is set to False in ttest function in NCL, and
subroutine tutest.for is used in Numeric Recipi. But
actually, when I compared the results of ttest in NCL
and NR, I found the results in NCL and NR just
corresponded across, so I guess there is something
wrong with ttest function in NCL.
As for correlation function, suppose there are two
samples x and y, when I calculated the correlation of
x leading y, I did it in NCL like
x_lead_y=esccr(y,x,maxlag), but actually, I found the
result was not I wanted, it was y leading x
correlation, I have compared the result in NCL with
other results that I made with fortran and GrADS, so I
guess there is also something wrong with esccr
function in NCL.
I hope someone can do something confirming my guess,
and then give feedbacks to NCL workshops.
Department of Atmospheric Sciences
Do You Yahoo!?
ncl-talk mailing list
ncl-talk AT ucar.edu
This archive was generated by hypermail 2b29 : Mon Apr 14 2003 - 21:06:47 MDT