Re: [ferret_users] variance up to a time-varying level

[Andrew Wittenberg <Andrew.Wittenberg@xxxxxxxx>]

Hi Bill,

You could use a masking expression to eliminate (set to missing) all 
points above the highest cloud top level, and then compute the variance of 
that masked field.  E.g.

let w_below = IF k[gk=w] LE cloud_top_lev THEN w
let w_var = w_below[i=1:`w,r=iend`@var,j=1:`w,r=jend`@var,k=@var]

Andrew

On Tue, 4 Apr 2006, Gustafson, William I wrote:

> I have come up with an algorithm that works at one time, but it fails 
> when applied to a multi-time dataset because I have to evaluate an 
> expression to a scalar to get it into the k=1:`lev`@var command. If 
> anybody can get around this problem, I would be grateful for your 
> assistance.
>
> Here is what I am doing right now...
>
> let cloudthresh = 0.00001
> let cloudmixr = qcloud+qice
> let event_mask = cloudmixr[k=@evnt:`cloudthresh`]   !Counts number of cloudy
> cells from the bottom of the domain
> let event_max = event_mask[k=@max]   !At each column, find out how many
> cloudy cells exist
> let event_max_mask = if event_mask eq event_max then 1   !Make a mask that
> goes to 1 at the top of the clouds for each col.
> let cloud_top_lev3d = event_max_mask[k=@loc:1]    !The level of the highest
> cloud top within each column
> let cloud_top_lev = cloud_top_lev3d[i=@max,j=@max]  !The single highest
> cloud top level in the domain
> let w_var =
> w[i=1:`w,return=iend`@var,j=1:`w,return=jend`@var,k=1:`cloud_top_lev`@var]
> !Variance of w up to cloud_top_lev