Problem writing out landmask to history file

[ekluzek]

I noticed this on the cesm2_3 branch as exact restart tests fail. But, it's an issue that we could have noted on the cesm2_1 branch as well. PIO is failing because _FillValue isn't properly set.

This bit of PIO code is failing (pio_darray.c):

    /* Check that if the user passed a fill value, it is correct. If
     * use_fill is false, then find_var_fillvalue will not end up
     * getting a fill value. */
    if (fillvalue && vdesc->use_fill)
        if (memcmp(fillvalue, vdesc->fillvalue, vdesc->pio_type_size))
            return pio_err(ios, file, PIO_EINVAL, __FILE__, __LINE__);

I think this just means that we need to explicitly set missing_value and _FillValue in the definition of history and restart variables. So I think it's a straightforward fix.

[ekluzek]

It looks like what's killing the model running is when it writes out landmask to the history file. landmask is defined as integer and has ifill_value and imissing_value set to ispval (and ispval = -9999) which should be valid. If I comment out the write of landmask it'll run to completion.

[ekluzek]

@marysa and @islasimpson I'm assuming that the solution of commenting out writing of landmask on the history is NOT a problem for you. Let me know if that's an important field for any history file processing you do.

I do think this should be fixed eventually, but it seems like dropping it for now should be fine. Let me know if I'm wrong about that.

[islasimpson]

I think it's somewhat important to have the landmask available but I guess people can obtain the landmask by other means, such as looking at the input files. So it seems like an ok solution to drop it for now.

[ekluzek]

Yes, landfrac is still output. And I think if you create a mask from it based on when landfrac==0 you'll get the landmask. Although I should verify this is the case so we know.

I also think this is a bug that was fixed in history for other models, so we likely just need to bring in the same fix from CTSM or MOSART. I think it's an issue with having an updated PIO version.

[marysa]

Yep, I agree with Isla - as long as there is some way of backing out "where does the model think there is land" from the history file, I don't think this is a problem.

…

On Mon, Jun 12, 2023 at 8:44 AM Erik Kluzek ***@***.***> wrote: Yes, landfrac is still output. And I think if you create a mask from it based on when landfrac==0 you'll get the landmask. Although I should verify this is the case so we know. I also think this is a bug that was fixed in history for other models, so we likely just need to bring in the same fix from CTSM or MOSART. I think it's an issue with having an updated PIO version. — Reply to this email directly, view it on GitHub <#82 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB3PPY36DUCEPIGZNJNFPQDXK4TLPANCNFSM6AAAAAAZB4GDMQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Marysa Laguë Adjunct Assistant Professor Department of Atmospheric Sciences University of Utah marysalague.com The University of Utah is located on the traditional and ancestral homeland of the Shoshone, Paiute, Goshute, and Ute Tribes. The University of Utah recognizes and respects the enduring relationship that exists between many Indigenous Peoples and their traditional homelands. We respect the sovereign relationship between tribes, states, and the federal government, and we affirm the University of Utah’s commitment to a partnership with Native Nations and Urban Indian communities through research, education, and community outreach activities.