Question about ERA5 invariant fields and soil moisture interpolation

[seti]

Dear all,


I performed two tests using exactly the same ERA5 meteorological data and WPS configuration.


In the first test, I ran metgrid.exe without including the two invariant fields. The resulting soil moisture field (SM000007, 0–7 cm) looks relatively smooth and spatially continuous.


In the second test, I separately ungribbed the ERA5 land–sea mask and surface geopotential and included the resulting constant file in metgrid.exe using:


constants_name = 'cons:2016-01-01_00'


In this case, the soil moisture field changes noticeably and appears much more blocky/patchy compared with the first case.


Interestingly, in the case without the invariant fields, I checked the soil moisture values over land and found no negative values and no unrealistic values above 0.5 m³ m⁻³. However, I am not sure whether the smoother field is actually the correct interpolation or whether it only looks more reasonable visually.


Could you please advise which approach is technically correct for ERA5–WPS preprocessing? Should the ERA5 land–sea mask and surface geopotential always be supplied as invariant fields to metgrid.exe, even when the soil moisture field obtained without them appears smoother.
Thanks in advance

 

[seti]

@Ming Chen

 

[seti]

And here is the way I download the data. I do not now if I downoload the right data, because the provided links in the froums are not available.

/FS/DECS/DS630.0/e5.oper.invariant/201601/e5.oper.invariant.128_172_lsm.regn320sc.2016010100_2016010100.grb (land-sea mask)
/FS/DECS/DS630.0/e5.oper.invariant/201601/e5.oper.invariant.128_129_z.regn320sc.2016010100_2016010100.grb (terrain geopotential, m2/s2)

 

[Ming Chen]

Hi Seti,

Thank you for the plots and detailed description of the issue.

In your first approach, you mentioned that two invariant fields are excluded, --- I guess you excluded soilght and landmask. Please let me if I am wrong.

My question is: Is there any special reason that you want to exclude these two fields?

Both approaches you described in your post are not correct.

We recommend that you include soilght and landmask in your intermediate files (i.e., the first approach you used but including the two fields).

 

[seti]

Dear Ming,

Thank you very much for your reply.


Yes, in my first approach I intentionally excluded the two invariant fields, namely the land-sea mask and the terrain/geopotential-related field, only as a sensitivity test. I wanted to compare how the soil moisture field would look with and without these two fields and understand how strongly they affect the interpolation in metgrid.


When I excluded them, the resulting soil moisture field looked much smoother and more spatially reasonable. However, when I included both invariant fields, the soil moisture field became much more blocky and patchy.


I should also clarify that when I included the invariant fields, I used both the land-sea mask and the surface geopotential/terrain-height information.


After that, I downloaded the ERA5 surface data again directly from the Copernicus Climate Data Store and checked the intermediate files carefully. I found that the surface dataset itself already contains the land-sea mask, source terrain height,.


I then rebuilt the intermediate files and reran metgrid using these ERA5 surface fields, without adding the separate invariant file again. The resulting soil moisture field was again very similar to the smoother case obtained when I had excluded the separate invariant file.


So now I am a bit uncertain about the correct workflow.


Should I use the land-sea mask and source terrain height that are already present in the ERA5 surface intermediate files, or should I still add the separate invariant file through constants_name?


In my current surface intermediate files, both LANDSEA and SOILHGT are already present together with the soil moisture fields.


Could you please advise which approach is correct in this case, and whether the separate invariant file is still necessary?


Thank you very much for your help.


Kind regards,