Good afternoon, everyone!

I'm running a coupled OpenIFS 43r3 + NEMO ORCA05 model with a runoff mapper similar to that of EC-Earth.
The runoff mapper collects surface and sub-surface runoff from HTESSEL and eventually sends this as runoff to NEMO.

However, I noticed that the "runoff" in NEMO (actually calving) coming from Antarctica was zero during the austral winter.
This leads me to the question: How is runoff actually computed over Antarctica?

I did a 1 year simulation and plotted total E+ES+TP (precipitation minus evaporation) and RNF (surface runoff + subsurface runoff). The numbers are roughly mm/year.
The output has been interpolated from the reduced O96 grid to a regular grid so the numbers are not exactly what OpenIFS uses internally.
I find plenty of net downward freshwater fluxes over Antarctica (over 100 mm/year) that does not correspond to any runoff.
This only makes sense if freshwater is advected central Antarctica towards the ocean by HTESSEL.

Is the water flux over this region transported towards the ocean by HTESSEL, and then accumulated in the runoff field?
Or does HTESSEL ignore snowfall over this region, i.e. the freshwater budget is not closed?

Many thanks
Joakim


5 Comments

  1. Unknown User (nagc)

    Hi Joakim, that's not a question I can answer. I'll find someone who can.

    Glenn

  2. Unknown User (nagc)

    Here's the response I've received:

    "In short the runoff over Antarctica (and by analogy of treatment also over Greenland) is not really useable at face value because we do not have a real mass balance over ice-sheet and glaciers (with the 10m of snow cover). A climatology might do better. If this gets used anyway it may provide too little runoff."

  3. Unknown User (de3j)

    Hi Glenn

    Thanks for digging this up. After recommendation from Jan Streffing I added this to my "runoff" sent from OpenIFS to the coupler:

    ZEXSNOW(IL:IL+IE) = MAX(PSURF%PSP_SG(IL:IL+IE,YSP_SG%YF%MP9)+ PSURF%PSNSE1(IL:IL+IE)*TSTEP-10000.0_JPRB,0.)

PSURF%PSNSE1(IL:IL+IE) = PSURF%PSNSE1(IL:IL+IE) - ZEXSNOW(IL:IL+IE)/TSTEP

CPLNG_FLD(CPLNG_IDX('A_Calving'))%D(IG:IG+IE,1,1) = (ZEXSNOW(IL:IL+IE)/TSTEP)

    This takes all snow that falls where snow depth > 10m, adds it to a new field "Calving" which is sent to the coupler, and also removes the tendency from OpenIFS.

    I will have to run some tests to see if this improves water balance in the coupled system and I'll report back later.

    Another option, I guess, could be to take all P-E over all land ice and use as "calving"?

    Best wishes
    Joakim

  4. Thomas Rackow

    In order to see whether your Calving makes sense, you could also compare to climatology estimates I once produced: https://doi.pangaea.de/10.1594/PANGAEA.865335?format=html#download (if this is useful at all).

    These have been scaled to an annual-mean input of 1321 Gt per year (from calving), following the estimate by Depoorter et al.

  5. Unknown User (de3j)

    Hello again

    I've dug into this issue a bit more and investigated the P + E - R balance (total precip + evap - surface runoff - subsurface runoff).
    As I see it, this should be zero, unless there is some accumulation of water in the soil. Over a long enough time this should be near zero.

    The plot below shows that P + E - R is approximately zero over most of the land, but clearly not over Greenland and Antarctica.
    Also odd is the time series of soil water over Antarctica which is mostly a constant value except for some sudden drops every few decades.

    It seems that the precip over Greenland and Antartica is not being "transformed" into runoff properly, and hence the model is missing freshwater.
    I also find that Antarctic runoff is lower in my climate runs compared to climatologies from e.g. JRA55.

    A possible solution could be to explicitly set R = P + E over ice sheets.
    Is there a way in OpenIFS to identify Greenland and Antarctica to do this?

    Or is there another way to ensure conservation over ice sheets?

    Many thanks
    Joakim