[MITgcm-support] filters for vertical velocity in cubed-sphere grid

[Andrea Molod]

hi yuan,

> I noticed Shapiro filter is not applied to vertical velocity (atmospheric 
> model, maybe generally for oceanic model as well).
> Ideally, the vertical velocity should be calculated from continuity equation. 
> However, corner effect on vertical velocity are
> noticeable, ie, stronger upward or downward motions comparing to elsewhere. 
> In a neutrally stable atmosphere, air at corners
> will experience adiabatic heating or cooling and cause local maximum (or 
> minimum) of temperature.

can you tell me which setup/experiment etc... you are using? i ask
because there are a set of grid files for the 32x32 cube that were put
together by jean-michel to address this issue. indeed, the vertical
velocity at the corners under certain circumstances got noisy, and
he made some adjustments to the locations of the grid cell edges
relative to the centers (if i have this correct...) that made most
of this go away. you can find them in the directory: 
verification/fizhi-cs-32x32x40/input and they are called dxC1....

as far as i understand things, trying to put the vertical velocity
through the shapiro filter would not 'really' address the problem,
it would sort of mask it and maybe even make a bigger mess.

>
> Also, it there any way to generate high resolution cubed-sphere grid while 
> maintaining the ratio of grid areas?
>

again, as dimitris asked, what do you mean by 'high res'?
(the link he gave you is for the 'really' high res - i have generated
a new cubed sphere grid FROM the one he talked about with just under 1 
degree resolution). also - as far as i know, any of the cubed sphere
grids that people here have generated try to maintain the good ratio of 
grid areas.....

hope some of this answers your questions.

andrea