Extreme cold bias - not snow cover

[dani_caputi]

Hello NCAR team,

I have been running forecasts out to 48h using the GFS model for my boundary conditions, allowing 24h of spin-up. Looking at a particular set of my runs over the Milwaukee, WI region from 10/24 - 11/3 (I was using the wxchallenge as a test for my project), I have an average 11 K cold bias for daily high temperature and 7 K cold bias for daily low temperature. This bias was similar in magnitude in all regions I tested it, and there was no snow cover that time of year. I don't think it is an urban physics issue either, because the bias is similar across the entire area of the model that includes both urban and rural regions.

I've attached my namelist.input. Also, as a sanity check - my way of calculating temperature from the wrfout_* netcdf files is to take the base temperature and add the perturbation temperature. In python this would be:

from netCDF4 import Dataset
wrf = Dataset('wrfout_xxx'), 'r')
T0 = wrf.variables['T00'][0] #stores base temperature as T0
T = wrf.variables['T'] #stores perturbation temperatures as T
Surface_Temp = T[0,0,x,y]+T0 #calculates surface temperature at grid cell x,y

 

[haiqingsong]

Hello NCAR team,

I have been running forecasts out to 48h using the GFS model for my boundary conditions, allowing 24h of spin-up. Looking at a particular set of my runs over the Milwaukee, WI region from 10/24 - 11/3 (I was using the wxchallenge as a test for my project), I have an average 11 K cold bias for daily high temperature and 7 K cold bias for daily low temperature. This bias was similar in magnitude in all regions I tested it, and there was no snow cover that time of year. I don't think it is an urban physics issue either, because the bias is similar across the entire area of the model that includes both urban and rural regions.

I've attached my namelist.input. Also, as a sanity check - my way of calculating temperature from the wrfout_* netcdf files is to take the base temperature and add the perturbation temperature. In python this would be:

from netCDF4 import Dataset
wrf = Dataset('wrfout_xxx'), 'r')
T0 = wrf.variables['T00'][0] #stores base temperature as T0
T = wrf.variables['T'] #stores perturbation temperatures as T
Surface_Temp = T[0,0,x,y]+T0 #calculates surface temperature at grid cell x,y

What version of WRF do you use?

 

[dani_caputi]

I cloned the latest from GitHub on 9/18/2022, so I believe 4.4.1

 

[haiqingsong]

I cloned the latest from GitHub on 9/18/2022, so I believe 4.4.1

This may help you！ Positive temperature bias in results after switching to WRF 4.4 using NoahMP · Issue #1747 · wrf-model/WRF

 

[dani_caputi]

This may help you！ Positive temperature bias in results after switching to WRF 4.4 using NoahMP · Issue #1747 · wrf-model/WRF

I'm a little confused as to the solution being proposed here. This is suggesting I use Noah instead of Noah-MP, if I am understanding correctly? In my namelist, sf_surface_physics = 2, so I believe this is already set to Noah rather than Noah-MP.

 

[eck.naresh]

I'm a little confused as to the solution being proposed here. This is suggesting I use Noah instead of Noah-MP, RTDC ki sthapana

 

[Ming Chen]

Have you looked at T2, which can be extracted from wrfout directly?

By the way, the potential temperature should be calculated by THETA = TT + 300

It is not correct to use T00.

 

[dani_caputi]

Have you looked at T2, which can be extracted from wrfout directly?

By the way, the potential temperature should be calculated by THETA = TT + 300

It is not correct to use T00.

Thank you for the corrections! I somehow missed T2 as an output - the numbers are a lot more reasonable now.

I don't see 'TT' as a variable but I assume you mean 'T' - "perturbation potential temperature theta-t0"

 

[Ming Chen]

That is correct, Sorry for the typo.