[ncl-talk] inconsistent results of taylor_stats

[Beáta Szabó-Takács]

Dear Ekin,


Thank you very much for your response. It is now much more clear for me.


Kind regards,

Beáta

________________________________
From: ncl-talk-bounces at ucar.edu <ncl-talk-bounces at ucar.edu> on behalf of Ekin Akoglu <ekin at ims.metu.edu.tr>
Sent: Thursday, August 3, 2017 10:30:43 AM
To: ncl-talk at ucar.edu
Subject: Re: [ncl-talk] inconsistent results of taylor_stats


Dear Beata,

I haven't gone through your results in detail but what I understood from your description of the problem, in my humble opinion, the results of taylor_stats may not be wrong. The correlation values just show that your model simulation results increase and decrease in agreement with / parallel to your data, i.e. simulation results increase/decrease when there is an increase/decrease in the data. However, correlation is not a measure of goodness of fit, a.k.a model skill. Bias shows how much your model over/underestimates the data. Hence, you may have a high correlation with a low bias as well as a high correlation with a high bias.

Best regards,

Ekin

On 03/08/17 11:18, Beáta Szabó-Takács wrote:
Dear NCL users,
I have doubts with taylor_stats functions. I created taylor diagrams from five RCMs in eigth sub area and also created tables about absolute bias of temperature and precipitation. I attached the diagrams in pdfs and the resulted bias values in .png. According to the taylor diagrams the temperature has large correlation (0.9 or larger) whilst the precipitation has smaller correclation. But the bias is larger in the case of temperature than in the case of precipitation. I do not understand why I got inconsistent results. I attached the sript of taylor diagram (taylor_ENS.ncl) and the sripts of bias tables (prtaylor2.ncl and temptaylor2.ncl) and the taylor_stats.ncl. Can someone inform me what I did wrong?
Thank you very much for your response in advance!
Kind regards,
Beáta




_______________________________________________
ncl-talk mailing list
ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk



--
Ekin Akoglu, PhD

Assistant Professor
Institute of Marine Sciences
Middle East Technical University
P.O. Box 28, 33731
Erdemli, Turkey

Email: ekin at ims.metu.edu.tr<mailto:ekin at ims.metu.edu.tr> / eakoglu at metu.edu.tr<mailto:eakoglu at metu.edu.tr>
Phone: +90 324 521 24 06 / Ext. 1328
GSM: +90 506 554 03 90
Skype: ekinakoglu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20170803/6340c215/attachment.html