Skin Temperature issue from ERA5

[htan2013]

Dear experts,

I tried to run the model with 240 m resolution over the Chicago region. The 240m domain is the innermost domain where the parent domains are 720m and 2160m. I used ndown to process each domain and the IC&BC are coming from ERA5. Because the resolution that I used is relatively higher than ERA5, I used another SST data (1.3km resolution) to replace the default SST in wrflowinp (see attached #1 & 2).

After the simulation finished, I found out that the T2 and TSK are not quite correct over the lake region (see attached #3&4) although the SST in wrfout seems to be correct (attached #5). I have checked the landuse (#6) which is correct. I am wondering:

1) How do we make TSK/T2 correct? I believe T2 is from TSK and TSK comes from input data (ERA5, #8).
2) I am thinking about copying the updated SST into TSK. However, due to the resolution difference, if I do that, I will still see a non-smooth, irregular transition boundary.
3) Is smooth_option going to work in this case? The innermost domain's input here comes from the parent domain by ndown.

 

[htan2013]

Another issue that I found from ERA5 input data (Downloaded from Copernicus) is the landmask. Apparently, the landmask is all 1 which is not correct. I did use script download ERA5 and did not incorporate ERA5_INVARIANT during metgrid because I have downloaded land_sea_mask...

 

[htan2013]

Another issue that I found from ERA5 input data (Downloaded from Copernicus) is the landmask. Apparently, the landmask is all 1 which is not correct. I did use script download ERA5 and did not incorporate ERA5_INVARIANT during metgrid because I have downloaded land_sea_mask...

The landmask in geo_em and met_em file is fine but the landmask in wrfinput is all 1..

 

[kwerner]

Apologies for the delay in response while our team tended to time-sensitive obligations. Thank you for your patience.

1) You mention that you downloaded your own LANDSEA mask, and therefore didn't use the ERA5_INVARIANT file during metgrid. What happens if you DO use the ERA5_INVARIANT file, instead of your version of the LANDSEA data?

2) Can you attach your namelist.input file?

3) Will you please let me know the version of WRF you're using?

4) Can you send a plot of LANDMASK in the met_em* file that corresponds to the same for the wrfinput* file you attached above?

 

[htan2013]

Apologies for the delay in response while our team tended to time-sensitive obligations. Thank you for your patience.

1) You mention that you downloaded your own LANDSEA mask, and therefore didn't use the ERA5_INVARIANT file during metgrid. What happens if you DO use the ERA5_INVARIANT file, instead of your version of the LANDSEA data?

2) Can you attach your namelist.input file?

3) Will you please let me know the version of WRF you're using?

4) Can you send a plot of LANDMASK in the met_em* file that corresponds to the same for the wrfinput* file you attached above?

Hi, thank you so much for your reply.

1) Yes, I have tried using ERA5_INVARIANT. However, when I tried metgrid, I met "ERROR: Cannot combine time-independent data with time-dependent data for field LANDSEA.mask". I am sure that I have land-sea mask variable in the ERA5 grib data.

2) The namelist for domain 3 is attached to the top. The namelist is for all domains for generating wrfinput is attached here.

3) If you see my namelist, the namelist has 4 domains. Right now I am using ndown to process each domain. For simulation in domain 1, no TSK error was found. However, this error starts to show after I process wrfbdy and wrfinput by ndown and run domain 2-3. I am not sure why but I do not see this error when I use HRRR data. The plots above are all for domain 2.

 

[htan2013]

Apologies for the delay in response while our team tended to time-sensitive obligations. Thank you for your patience.

1) You mention that you downloaded your own LANDSEA mask, and therefore didn't use the ERA5_INVARIANT file during metgrid. What happens if you DO use the ERA5_INVARIANT file, instead of your version of the LANDSEA data?

2) Can you attach your namelist.input file?

3) Will you please let me know the version of WRF you're using?

4) Can you send a plot of LANDMASK in the met_em* file that corresponds to the same for the wrfinput* file you attached above?

And also, based on my previous experience using ERA5 on WRF, the TSK in wrfinput_d01 always looks this way (those boxes). However, it should disappear after a couple of hours of simulation but it didn't.

 

[kwerner]

3) Will you please let me know the version of WRF you're using?

Thanks for the responses. Will you also answer this question from my above post, as well? Thanks!

 

[htan2013]

Thanks for the responses. Will you also answer this question from my above post, as well? Thanks!

Oh right, I am using WRF4.5.1.

 

[htan2013]

Thanks for the responses. Will you also answer this question from my above post, as well? Thanks!

Hi Kelly,

Do you have any updates on this issue? Several of my colleagues also have this issue when using ERA5 with multiple domains. The attached figure shows the 2-m temperature. We downloaded the data from Copernicus since the RDA no longer supports ERA5 for WRF.

 

[kwerner]

Sorry for the delay. I wanted to converse with some colleagues about this. We decided that instead of trying to figure out where something went wrong in the procedure you used to convert the ERA5 data, it may be easier to use this new conversion script we just released. See Converting ERA5 Model-level Data from netCDF to WPS Intermediate Format.