EOFs and vertical levels
Added by Fabian Paul over 13 years ago
Dear all
I'm tinkering with CDO's new EOF functions. I tried to decompose the global wind velocity field into principal components. But I noticed that CDO seems to perform the principal component analysis separately for each level. Is there any way to compute EOFs from the full 3d data set?
I did roughly:
cdo -v mul area_weighting.grb -sub chi3_motion.grb -timmean chi3_motion.grb chi3_motion_anom.grb
cdo -v eof,100 chi3_motion_anom.grb chi3_motion_ev.grb chi3_motion_eof.grb
cdo -v eofcoeff chi3_motion_eof.grb chi3_motion_anom.grb chi3_motion_pc
inspecting the eigenvector file yields:
fab@strat09:/scratch/users/fab/eams$ cdo showlevel chi3_motion_ev.grb
20000 25000 30000 40000 50000 60000 70000 77500 85000 92500 100000
cdo showlevel: Processed 1 variable. ( 0.00s )
fab@strat09:/scratch/users/fab/eams$ cdo showtime chi3_motion_ev.grb
00:00:00 [...]
cdo showtime: Processed 1 variable over 276 timesteps. ( 0.00s )
Thanks in advance
Fabian
Replies (1)
RE: EOFs and vertical levels - Added by Cedrick Ansorge over 13 years ago
Dear Fabian,
I am very sorry to tell you that there is no way to compute a full 3D EOF with the CDOs so far. However, the introduction of a 3D EOF operator is planned for the future.
Regarding your issue we will modify the documentation of the EOF operator to clearly state that it only computes 2D EOFs in order to avoid confusion in the future.
Regards,
Cedrick