Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Confusing values for 2m Temperature (T2), Potential temperature (2m) and heat flux (HFX)


New member
I was making 2 sensitivity WRF simulations over ICE COVERED OCEAN. The focus was on atmosphere-ocean boundary layer.

Following details/parameterizations were kept same for both simulations:
The duration of simulation was 2 days.
There were 4 domains nested one over the other. Resolutions of 4th and 3rd domains are 80m and 240m respectively.
Physics parameterizations used are same for both simulations.
The only difference between 2 simulations is that the input data for SEAICE concentration was modified.

In the first sensitivity simulation, inside domain 4, the SEAICE concentration (input data) values were either 1 (which mean ice) or 0 (which means open water). There are no intermediate values used (eg: 0.2, 0.3 etc). This means, it is either fully ice covered grid, or it is fully open water region.
In the second simulation, the SEAICE concentration values (input data) were uniform for entire domain 4. The value was 0.3 for all grid points. This means the grid is neither fully open water, nor is it fully ice covered.

I was successful in generating 10m-winds with reasonable accuracy in both cases.

However, in the first simulation, the 2m temperature (T2) and potential temperature at 2m (TH2) over sea ice grids were higher than the surrounding open water grids. According to my understanding, both T2 and TH2 are dependent on the surface temperatures. Now, considering that the surface temperatures over SEAICE is lower than that over the open water, the values of T2 and TH2 should've been lower over SEAICE grids.

Another concern I've is that, the model estimated heat fluxes (HFX) were positive values for the second simulation. The postive sign is expected to represent the direction, and positive means upward direction. However, in the first simulation, HFX values ranged in negative values irrespective of whether it is over SEAICE or over open water.

Why am I seeing such behaviours? Can someone please address this .

Few time averaged spatial (domain 4) figures are attached for reference. For each figure, the left hand side show the first simualtion and right hand side show the second simulation.

Here is the physics parameterization that I used:
physics_suite = 'CONUS'
mp_physics = -1, -1, -1, -1,
cu_physics = 0, 0, 0, 0,
ra_lw_physics = -1, -1, -1, -1,
ra_sw_physics = -1, -1, -1, -1,
bl_pbl_physics = 4, 4, 4, 0,
sf_sfclay_physics = 4, 4, 4, 4,
sf_surface_physics = 4, 4, 4, 4,

radt = 10, 10, 10, 10,
bldt = 0, 0, 0, 0,
cudt = 5, 5, 5, 0,
icloud = 1,
num_land_cat = 21,
sf_urban_physics = 0, 0, 0, 0,
num_soil_layers = 4,
surface_input_source = 1,
fractional_seaice = 1,
seaice_thickness_opt = 0,
seaice_thickness_default = 1.2,
seaice_albedo_opt = 1,
seaice_snowdepth_opt = 0,
seaice_snowdepth_min = 0,
seaice_snowdepth_max = 2,
tice2tsk_if2cold = .true.,
isfflx = 1,
ifsnow = 0,

Thank you in advance.

Hi everyone!
I think have resolved the concern. So, as I described in the concern, values for t2, th2, tsk were higher over the ice, than over the water. This is wrong and for a correct simulation, the results were supposed to happen the other way around.
Now, the reason for this behaviour is that, ice and open water grids in WRF simulations are not treated based on "SEAICE" alone. There are other parameters which decide these. Values of these parameters are also to be modified for a correct simulation. These parameters are - LANDMASK, LU_INDEX, IVGTYP, ISLTYP, TSK, SNOALB, TMN, XLAND.
In my case, I had modified only the SEAICE data initially, which lead to a wrong result. So, all other parameters (listed above) considered the domain as ice covered region only. This lead to improper calculation of fluxes and near surface temperatures.
Thanks for the update. Would you please clarify in detail what variables (seaice, landmask, lu_index, etc.) in which files (met_em, wrfinput) did you modify? I ask this because I think that REAL and WRF will automatically adjust variables like lu_index, ivgtyp, xland etc. once the open sea water point changes to seaice point and vice versa. However, such adjustments seemed not being conducted in your case.
Sorry for the lack of clarity. Like you said, the met_em files are processed by REAL.exe, so if I make a modification in "met_em" files, it could make necessary changes in the input files (wrfinput and wrfbdy).
However, in my case, I made changes to "wrfinput" file, which is the output file from running REAL.exe. This resulted in issues with specifying sea water and seaice points.
Thanks for your clarification. That makes sense because wrfinput is produced after REAL is run.