[CF-metadata] weighted time mean vs. conditional time mean.

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 16 Mar 2016 14:13:32 +0000

Dear Martin

> (1) Grid cells have a temporally varying sea-ice fraction, and a mean is wanted over the sea-ice area.
>
> Is this a correct cell_methods formulation:
> "time: mean where sea_ice area: mean where sea_ice"?

The where construction is only provided in CF1.6 for area fractions, not time
fractions (section CF 7.3.3). I think the method is first to average at each
time over the sea ice area, and then make a time mean. The second operation
does not have to know about the sea ice, so the cell_methods is

area: mean where sea_ice time: mean

> (2) Grid cells which sometimes have convective cloud, and a mean is wanted over the times when there is a cloud in the cell. The variable to be averaged is, in this case, the minimum over the cell of the convective cloud base.
>
> Is this the correct cell_methods:
> "time: mean where convective_cloud area: minimum"?
>
> In the first case the first case "mean where sea_ice" refers to the portion of grid cells have sea_ice, in the 2nd "mean where convective_cloud" would refer to whole cells ... is this OK?

Again, where is not for times. Also convective_cloud is not in the area-type
table, though it could be added I expect. I think I'd put

area: minimum time: mean (when convective cloud present)

where the parenthesis indicates an optional unstandardised comment.

Best wishes

Jonathan
Received on Wed Mar 16 2016 - 08:13:32 GMT