[MITgcm-support] exf interp for smooth wind stress curl

[Menemenlis, Dimitris (3248)]

Chris, thank you for getting back.

Embarrassingly, our atmospheric grid edge effects appear to have been artifacts of sloppy curl computation.  After repeating the computation in a way consistent with model grid, the edge effects went away.

Matt, I wonder whether the wind stress curl artifacts in your CASE (1/16^o) setup could originate from some other cause than the exf_interp bicubic interpolation?

For support list reference I attach the curl recipe that we used:

Wind stress curl at  [ XG(i,j), YG(i,j) ] = dV/dx - dU/dy
U(i,j) = EXFtaux(i,j)
V(i,j) = EXFtauy(i,j)
dV/dx = [ V(i,j) - V(i-1,j) ] / dxV(i,j)
dU/dy = [ U(i,j) - U(i,j-1) ] / dyU(i,j)

On May 18, 2011, at 7:55 AM, Chris Hill wrote:

> Last I recall J-M and I were waiting to see Matt's code - since the
> edges he had did not look entirely consistent
> with interpolation error.
> 
> Chris
> 
> On Wed, May 18, 2011 at 1:20 AM, Matthew Mazloff <mmazloff at ucsd.edu> wrote:
>> nope -- never tried Chris' superduper fix -- or maybe he has a higher
>> resolution atmospheric reanalysis hidden in the closet?
>> 
>> -Matt
>> 
>> 
>> On May 17, 2011, at 4:45 PM, Menemenlis, Dimitris (3248) wrote:
>> 
>>> Yes, we have artifacts at edges of atmospheric forcing grid.
>>> 
>>> So Chris never got back to you with his superduper offline fix for this
>>> problem?
>>> 
>>> Dimitris Menemenlis
>>> 
>>> On May 17, 2011, at 3:58 PM, Matthew Mazloff wrote:
>>> 
>>>> Hi Dimitris,
>>>> 
>>>> I do use  bicubic interpolation on all NCEP fields -- but I still have
>>>> some artifacts in my wind stress curl for my CASE (1/16^o) setup.  Is
>>>> this what you are finding?
>>>> 
>>>> -Matt
>>>> 
>>>> On May 17, 2011, at 3:24 PM, Menemenlis, Dimitris (3248) wrote:
>>>> 
>>>>> Chris and Matt, did issue below regarding bicubic exf_interp ever
>>>>> get resolved?
>>>>> 
>>>>> Is there a way to get smooth wind stress curl for hi-res
>>>>> integrations from coarse-res atmospheric forcing?
>>>>> 
>>>>> Dimitris
>>>>> 
>>>>> http://mitgcm.org/pipermail/mitgcm-support/2009-October/006305.html
>>>>> 
>>>>>> Hi matt
>>>>>> 
>>>>>> can you put your wind fields, grid etc... Somewhere @ MIT. I have
>>>>>> code
>>>>>> to fix this offline, but have never found a real setup where it
>>>>>> mattered!
>>>>>> 
>>>>>> Chris