[ncl-talk] Some comments are needed in mjoclivar_14.ncl on the MJO diagnostics page [SEC=UNCLASSIFIED]

[Eun-Pa Lim]

?Dear NCL developers


My colleagues Harry Hendon and Abhik  Santra at the Bureau of Meteorology want to report an issue regarding the computation of the RMM1 and RMM2 time series (eof_ts_cdata) and the expansion coefficients of olr, u200 and u850 (ceof_ts) in mjoclivar_14.ncl.

?> eof_ts_cdata = eofunc_ts(cdata,eof_cdata,False)

Harry and Abhik were trying to replace cdata with forecast data of a particular year to see if forecast amplitudes of RMM1 and RMM2 are realistic or not.

Because eofunc_ts_Wrap removes the time mean of the input data by default (i.e. 1-year time mean of forecast data for their case), their forecast MJO amplitudes were systematically biased.


This problem will occur when people try to compute the MJO amplitues with a subset of a dataset.


Could you please add a comment where eof_ts_cdata and ceof_ts appear in mjoclivar_14.ncl that users should compute the expansion coefficients manually rather than using eofunc_ts_Wrap if the removal of the time mean of the input data is not intended/desirable?


Please let me know if what I've explained is not clear.


Many thanks!


Eun-Pa


Research Scientist

The Australian Bureau of Meteorology

GPO Box 1289, Melbourne, VIC 3000, AUS

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20170725/8b8bb3ac/attachment.html