Hi, NCl experts
I found a problem when I made EOF analysis with NCL eof function.
The following is my ncl programm with old version:
***********************************************************
load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/contributed.ncl"
begin
f = addfile("sst_50-03_original.nc","r")
ss =f->sano
;begin eof via co-variance matrix
neval = 4
eof = eofcov(ss,neval)
pcvar = eof@pcvar
print(pcvar)
end
************************************************************
output results:
Variable: pcvar
Type: float
Total Size: 16 bytes
4 values
Number of Dimensions: 1
Dimensions and sizes: [4]
Coordinates:
(0) 26.49188
(1) 15.10827
(2) 8.423062
(3) 6.310274
______________________________________________________________________
My programm with new version is as follows:
***********************************************************************
load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/contributed.ncl"
begin
f = addfile("sst_50-03_original.nc","r")
ss =f->sano
;begin eof via co-variance matrix
neval = 4
eof = eofunc(ss,neval,False)
pcvar = eof@pcvar
print(pcvar)
end
**********************************************************************
result is:
Variable: pcvar
Type: float
Total Size: 16 bytes
4 values
Number of Dimensions: 1
Dimensions and sizes: [4]
Coordinates:
(0) 23.71038
(1) 16.73346
(2) 7.300104
(3) 7.019832
___________________________________________________________________________
Their results are different, which influence my analysis because if old version is right the first 2 modes are significant by North test (1982), however if new version is right, the first 2 modes are not significant.
Could you help me? I prefered new version because the computing speed is very fast for large matrix. And many of my results had been based on the new version. But the result of old version seem more reasonable in physics( same with fortran programm result).
If you need this nc dataset, I can transfer to you.
Looking forward to your answer!
Thanks a lot!
Jing
_______________________________________________
ncl-talk mailing list
ncl-talk@ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
This archive was generated by hypermail 2b29 : Wed Mar 02 2005 - 09:55:32 MST