Problem with tmc_layt

[maignan]

In thermosoil_humlev, we have:

       ! The deep layers in thermosoil where hydro is not existing
       DO jd = nslm+1, ngrnd
          shum_ngrnd_perma(ji,jd) = shumdiag_perma(ji,nslm)
          mc_layt(ji,jd) = mc_layh(ji,nslm)
          mcl_layt(ji,jd) = mcl_layh(ji,nslm)
          tmc_layt(ji,jd) = tmc_layh(ji,nslm)
       ENDDO

However we cannot have at the same time the volumetric water content (mc) and the total water content (tmc) equal to those of the nslm layer because the layer depths are varying.

We propose to remove the tmc_layt(ji,jd) line.

Where tmc_layt is used later in thermosoil_getdiff to compute the paca_tmp/pcapa variables, we propose to replace:

tmc_layt(ji,jg) / mille / dlt(jg)

with:

mc_layt(ji,jg)

Reported by Dan Zhu.

[jgipsl]

After discussion with Dan and Fabienne, we agreed to change :

tmc_layt(ji,jd) = tmc_layh(ji,nslm)
to 
tmc_layt(ji,jd) = tmc_layh(ji,nslm)/dlt(nslm) *dlt(jd)

in  thermosoil.f90 (Line 1393 of  trunk version: rev 4594.)

This solves the problems regrding the variation of soil layer depth (for temperature) below 2m.

Mail from Fuxing Wang, 11/09/2017

[jgipsl]

Done in trunk rev [4619]