Problem with running MPAS with ERA Interim model level files

[wcheng]

Hi:

I have a problem with running MPAS with ERA interim model level files. I followed the procedure in

http://metclim.ucd.ie/2018/03/wrfv3-9-1-with-era-interim-data-on-orr2/

to get the WPS intermediate file for pressure, RH, and GHT in addition to using ungrib.exe to get the intermediate
files for the 2D surface fields and 3D fields for u, v, T, SPFH. I "cat" all three files into one and then executed

./init_atmosphere_model

I am running for a case initialized on 2018-07-30_00 and with the "40962" grid. For some of the ice points in the polar region, I am getting very low theta at the 1st model level in x1.40962.init.nc. e.g., a value of 138.313 K at 72.553818 S 165.703690 E. I did not see very cold temperature at the 1st model level in the original ERA interim file. The 2m temperature and soil temperature in MPAS did not have temperatures that were that cold.

I used the same file to run WRF over Antarctica at the problematic point, the initial theta was around 220-230 K.

I am attaching my namelist and the theta at 1st level from x1.40962.init.nc.

I am using MODIS landtype (USGS had the similar problem).

Thanks in advance for any help you can offer!

 

[mgduda]

It's possible that the default vertical extrapolation that we use is to blame. If the MPAS terrain is significantly lower than the ERA-I terrain and if there's a surface inversion, the downward extrapolation of temperatures following the linear trend of the lowest two ERA-I points could lead to very cold temperatures. There is a namelist option that may help to address these issues: "config_extrap_airtemp". Could you try the following setting in your namelist.init_atmosphere file to see if helps?

&interpolation_control
    config_extrap_airtemp = 'lapse-rate'
/

This will cause the code to use a -0.0065 K/m lapse rate when extrapolating temperatures.

 

[wcheng]

Thanks, Michael! Your suggestion did the trick.

 

[Geess_321]

Hi wcheng,
You mentioned "to get the WPS intermediate file for pressure, RH, and GHT in addition to using ungrib.exe to get the intermediate
files for the 2D surface fields and 3D fields for u, v, T, SPFH. I "cat" all three files into one and then executed".
Is it OK to use 'cat' command to combine these files?
Use EC file (seperate the surface and model level and pressure level) to drive MPAS-A, it seems my model level variable such as u, v are not recognized by MPAS. and the wind is very small compared with original EC.
Please give me some advices and thanks.@wcheng @mgduda

 

[mgduda]

Using the 'cat' command to combine intermediate files should be fine.

Can you explain why you suspect the model-level variables are not recognized by MPAS? If you could attach your log.init_atmosphere.0000.out file, that would be helpful.

 

[Geess_321]

Using the 'cat' command to combine intermediate files should be fine.

Can you explain why you suspect the model-level variables are not recognized by MPAS? If you could attach your log.init_atmosphere.0000.out file, that would be helpful.

Thanks for your reply.
I use ECMWF and WPS to create ML:2020-08-06_12~ML:2020-08-06_18 and SURFACE:2020-08-06_12~SURFACE:2020-08-06_12, also I use ecmwf_coeffs and calc_ecmwf_p.exe to create PRES:2020-08-06_12~PRES:2020-08-06_18. Then I cat ML:2020-08-06_12, SURFACE:2020-08-06_12 and PRES:2020-08-06_12 into one binary file SMPP:2020-08-06_12, loop for each time.
Then I use these intermediate files to create initial condition and lateral boundary conditions in limited-area MPAS, though there are initial condition files and LBCs files created without any error. However, values in the XXX.init.nc file are all Infinity (same as missing value).
By the way, I can use GFS data run without any issue.
My attachment files as follows url: https://github.com/Geess126/MPAS-issue-files

 

[mgduda]

It seems as though there may be some mismatch in the vertical levels in your ERA-Interim intermediate files. Your 'ecmwf_coeffs' file has 138 levels, while your log.init_atmosphere.0000.out_ec_init suggests that your PRESSURE and other 3-d fields have only a subset of these levels.