Hi

I've been running OpenIFS using either L60, L91 or L137 vertical levels, which are all standard configurations. 

But is it possible to configure my own vertical grid? Specifically, I'd like to increase the vertical resolution in the lower atmosphere. 

And if I increase the vertical resolution, do I need to alter some vertical diffusivity/viscosity coefficient or minimum mixing length somewhere? 


Cheers

Joakim 

3 Comments

  1. Unknown User (nagc)

    In principle yes you can set any vertical resolution you like. In practise this is not a trivial exercise. A great deal of work would need to be done to tune various parts of the model and validate it - this is months of work. I would not recommend it.

    Cheers,  Glenn

  2. Unknown User (de3j)

    Thanks for the reply, Glenn. I'll stay away from doing this then! Seems like a dark rabbit hole to go down... 

    Looking through the "vtables" directory, I've found a whole bunch of different vertical grid configurations. Will OpenIFS work well with any of these (and is it possible to generate initial conditions with prepIFS)? Are L19 and L31 still supported in OpenIFS?  

    We've got some simulations with ECHAM at Geomar running T159L19 and T159L37, and they show that L37 reduces some SST biases in the tropics over L19. I'd like to extend that by running T159L19, T159L31, T159L60 and T159L91, to see if the results are the same in OpenIFS, and if the biases can go down even further with really high vertical resolution (L91). 

    Many thanks!

    Joakim  

  3. Unknown User (nagc)

    The vertical configurations in the 'vtables' directory should work but I think some will be test configurations. I recommend staying with the operational resolutions or those used for reanalyses (Interim, ERA-5 etc).

    You can find the details of the main supported level configurations here: 4.4 OpenIFS: Vertical Resolution and Configurations in the OpenIFS user guide to check the resolution in the lower atmosphere.

    Regarding changing the number of levels, the key thing is to devise an algorithm to create the A & B coefficients to make sure there are smooth changes in pressure between the levels. This needs to be solved for typical pressures at the surface and over high orography. The vertical finite element scheme is sensitive to the pressure spacing.

    And also consider the ratio between horizontal and vertical resolution: https://journals.ametsoc.org/doi/10.1175/1520-0493%281989%29117%3C2575%3ACVAHR%3E2.0.CO%3B2

    Cheers,  Glenn