Re: [ferret_users] variance Calculation

[USHNANSHU DUTTA <ushnanshudutta@xxxxxxxxx>]

let me elaborate my query. The link you forwarded was impressive but I am still doubtful about my problem.

I have 10 years JJAS daily data i.e 10 x 122 =1220 time steps.

Now I need to calculate the variance for each year over a certain [lat x lon box].

So I expect that output will be of 10 timesteps. 

But through the method given in the link, I am getting the output of 1220 timesteps.

Please help.

On Thu, Oct 4, 2018 at 8:39 AM Ryo Furue <furue@xxxxxxxxxx> wrote:

On Wed, Oct 3, 2018 at 9:02 PM USHNANSHU DUTTA <ushnanshudutta@xxxxxxxxx> wrote:

Is it possible to compute variance for each year in a single scrpt in ferret?

Here Andrew shows a method to calculate a "running correlation":

https://www.pmel.noaa.gov/maillists/tmap/ferret_users/fu_2018/msg00306.html

The method includes (or implies) how to calculate the running variance.  (See my response to Andrew.)

The value you want is the running variance at the center of each year.  (I haven't thought about how to extend it to cases where the length of the year isn't constant.)

Hope this helps,

Ryo

--

Ushnanshu Dutta

Research Fellow

Indian Institute of Tropical Meteorology

Pune,411008

Mob: 9405022799