**To**:**ferret_users@xxxxxxxx****Subject**:**Re: R: Re: [ferret_users] Different results between ferret and other softwares****From**:**Ryo Furue <furue@xxxxxxxxxx>**- Date: Thu, 28 Jul 2011 09:24:00 -1000 (HST)
- In-reply-to: <7070506.3713401311876254748.JavaMail.defaultUser@defaultHost>
- References: <7070506.3713401311876254748.JavaMail.defaultUser@defaultHost>
- Sender: owner-ferret_users@xxxxxxxx

Hi Markus, | However, I have a doubt about the average to use: is it better to | use the weighted average or the non-weighted average to compute the | mean global temperature ? The weighted average is the average you want, unless you have specific reasons to use other types of average. You use numerical models, which are meant to model the reality. And in reality, the usual mean temperature is [integral d(area) T] / (total area). Ferret computes an approximation to this value using the gridded T field. The "weight" we are talking about is just a natural consequence of this formulation. Another, practical argument is that a properly-weighted average depends less on the underlying coordinate system or on the grid spacing. Suppose, for example, that you increase the latitudinal resolution near the equator, so that you have more gridpoints near the equator. Your unweighted-average will suddenly rise, a result which normally you don't want. Regards, Ryo

**References**:**R: Re: [ferret_users] Different results between ferret and other softwares***From:*markus.mingel@xxxxxxxxx

- Previous by thread:
**R: Re: [ferret_users] Different results between ferret and other softwares** - Next by thread:
**[ferret_users] Default nlevels**

Dept of Commerce / NOAA / OAR / PMEL / Ferret

Privacy Policy | Disclaimer | Accessibility Statement