Should hfx_sub be substracted to compute qt_oce_ai?

Hi!

We, at CCCma, are having a heat conservation issue in our coupling with out atmospheric model CanAM, and looking at the NEMO-SI3 side, we have a question:

In ice_thd_dh, where hfx_sub is computed, a comment says

!    comment: not counted in mass/heat exchange in iceupdate.F90 since this is an exchange with atm. (not ocean)
hfx_sub_1d    (ji) = hfx_sub_1d    (ji) + ze_s(ji,jk) * zdum * a_i_1d(ji) * r1_Dt_ice  ! Heat flux of snw that sublimates [W.m-2], < 0

but in ice_update_flx the heat transfered to the ocean is computed using this hfx_sub

         IF( ln_icethd ) THEN
            qt_oce_ai(ji,jj) = qt_atm_oi(ji,jj) - hfx_sum(ji,jj) - hfx_bom(ji,jj) - hfx_bog(ji,jj) &
               &                                - hfx_dif(ji,jj) - hfx_opw(ji,jj) - hfx_snw(ji,jj) &
               &                                + hfx_thd(ji,jj) + hfx_dyn(ji,jj) + hfx_res(ji,jj) &
               &                                + hfx_sub(ji,jj) - SUM( qevap_ice(ji,jj,:) * a_i_b(ji,jj,:) ) + hfx_spr(ji,jj)
         ENDIF

Also, I was wondering if the latent heat of fusion of ice and snow is not already removed from Qns in the atmospheric side of the coupler as Qns includes the latent heat fluxes (evaporation over the ocean and sublimation over sea ice) as it is done in IPSL-CM6 (https://forge.ipsl.jussieu.fr/igcmg/wiki/IPSLCM6/Coupling)

Thanks in advance!

Well, there may be a confusion here.

The latent heat of ice/snow fusion is included in the downward non solar flux qns_ice (and qns_tot). And if not, it should be. I think this part is well taken into account by the atmospheric models. So, SI3 takes this heat to sublimate. So far so good. But sublimation comes with an ice-atmo mass exchange as well. And the sublimated ice has a non zero temperature. Hence, there is a sensible heat associated with this sublimation that should be retrieved by the atmosphere but it is not the case. This flux is called qevap_ice. In practice, qevap_ice is always set to 0, which means assuming a sublimation at 0°C and no sensible flux related to sublimation.

In SI3 the sensible flux associated with sublimation cannot be 0 since the sublimated ice has a non zero temperature. This flux is diagnosed and called hfx_sub. We cannot send it into the atmosphere. Therefore, for conservation purpose, it is included in the non solar flux reaching the ocean qt_oce_ai whose expression is given below:

qt_oce_ai(ji,jj) = qt_atm_oi(ji,jj) - blabla
   &             + hfx_sub(ji,jj) - SUM( qevap_ice(ji,jj,:) * a_i_b(ji,jj,:) ) + hfx_spr(ji,jj)

You see that, in a “perfect world”, hfx_sub = qevap_ice, and no sensible heat related to sublimation is sent to the ocean.

Hi Clement,

thanks a lot for your answer,

I completely agree, we do include the latent heat fluxes of evaporation and sublimation in Qns_x.

We were more confused with the sensible and mass heat fluxes, and what is included in our coupling interface vs what is expected. Thanks for your explanation, it makes total sense, but I will check how sensible fluxes are taken into account in our model so we are consistent with NEMO (we do not use OASIS).