Problem in temperature at 2 meters

[kranke]

Hello,

I've run some real simulations of a deep convective boundary layer and I am encountering some issues with the diagnosed T2 variable - I was expecting T2 to be bounded between the ground temperature (TSK) and and the first model temperature (T evaluated at first model level), following Monin-Obukhov similarity theory, but it's actually lower than T evaluated at the first model level in the afternoon, which does not make any physical sense. This problem happens clearly with the MYNN surface layer scheme (option 5). I am using the unified Noah land-surface model (option 2). Is that a known issue of the surface layer scheme or am I missing something? Is the final T2 diagnosed by the surface layer scheme or by the land surface model (I tried to look in the code but it's overwritten quite a few times and it's hard to get a handle on what's the final value for T2)?

Thanks for any insights you may give!

Paolo

 

[Ming Chen]

Hi,
Would you please send me your namelist.input and namelist;wps to take a look? It also will be helpful to tell what data you use to drive WRF.
Thanks.

 

[kranke]

Thank you for your quick reply! I'm attaching my namelist.input and namelist.wps if that's helpful (originally those were nested simulations, but I set max_dom=1 in the namelists I'm attaching since the problem also happens in d01). As a reference, here's a quick plot of the diurnal cycle of the different temperatures - I was expecting the black line to be very close to the green one (the green is my a-posteriori rough calculations with MO similarity theory for negative Richardson numbers --> free convection), or at least bounded between the blue line and the red one.



The plot refers to one specific grid point in the domain (i=71,j=71), hourly output for July 25th 2016

Thanks again for any help you can give me!

Paolo

 

[Ming Chen]

Paolo,
Thanks for sending the namelist files. I looked at your namelist.input. There are a few issues I am concerned:
(1) I suppose this case was initialized at 00UTC 22 July, and ran for 9 days. I would suggest that you tun on sst_update since this is a relatively long run.
(2) Can you try rrtmg radiation scheme, and set radt =12?
(3) I notice that num_metgrid_levels = 26, and p_top =5000. Ny question here is, what data did you use to drive this case? It seems that the vertical resolution of this data is pretty coarse. I am not sure whether this may impact the results. if convenient, can you try a different forcing data such as GFS or ERA5?
(4) if you run with a different surface layer scheme, for example sf_sfclay_physics = 1, can you get the expected, reasonable T2 ?

 

[kranke]

Hi and thanks for your reply! I forgot to mention that the simulation is driven by high-resolution forecasts from ECMWF (https://www.ecmwf.int/en/forecasts/datasets/set-i). I will run it with different boundary conditions (GSF or ERA5) soon, as well as I will try the different other options you suggested.

In the meanwhile, I tried another simulation with MYJ PBL scheme in combination with sf_sfclay_physics = 2 and the problem does not appear (namelist attached). This is the same figure for the new simulation:



I checked at all times in the model and T2 is always bounded between TSK and the T at the first model level (as expected). Could it be an issue from the MYNN surface layer scheme then (maybe in combination with the specific radiation scheme and Land Surface Model?)
Thanks!
Paolo