Microphysics scheme questions

[jiskattema]

modbulkmicro_point.f90

• [x] l_sb_inuc_expl not implemented (line 465)
• [x] output of tendencies dq_ci_dep, dq_hs_dep. dq_hg_dep is after the cor_deposit3 correction, line 870
• [x] if not l_sb_lim_aggr, E_ab was unset? setting it to zero now. line 942
• [x] elseif statement should be an else. line 1367
• [x] if not (D_cl.gt.D_c_a).and.(D_hg.gt.D_i0), can we turn off coll_gcg3? line 1474
• [x] if not (D_cl.gt.D_c_a).and.(D_ci.gt.D_i0), can we turn off coll_ici3? line 1219
• [x] if not (D_cl.gt.D_c_a).and.(D_hs.gt.D_i0), can we turn off coll_scs3? line 1323
• [x] evapmelting3: threshold for melting snow, should that be q_hs.gt.qicemin? line 2229
• [x] evapmelting3: threshold for melting graupel, should that be q_hg.gt.qicemin? line 2247
• [x] accretion3: when not l_sb_classic, should accretion happen when q_hr < q_hr_min? line 2501
• [x] evap_rain3: is the use of unlimited x_hrf correct? line 2617
• [x] satadj3: limit on dn_cl_sa is this now correct? is used to include svp(in_cl) (to include tendencies from outside the scheme?). I now initialize n_clp from svp(in_cl). line 2718 and 2751
• [x] sb_evmelt3: line 2997 and 3000, should the limit also include the tendency? -n_cl/delt - n_clp
• [x] integrals_bulk3: do we need to limit x_hr? line 370 and 389

modbulkmicro_column.f90

• [x] nucleation3: original code went out of bound (memory corruption) for d ssat / dk . line 130. Is this better?
• [ ] line 360. Sedimentation threshold for rain
• [ ] line 528. Sedimentation threshold for snow
• [ ] line 726. Sedimentation threshold for ice
• [ ] line 831. Sedimentation threshold for cloud water
• [x] line 631, line 634. wrong variable (typo) be_hs => be_hg
• [x] lines 368 and 385, 412, Dvr_spl and x_hr are not calculated for different settings. Do we need to limit them, too?
• [ ] General sedimentation issue. The time-splitting scheme is to prevent stuff falling faster than 1 k-level. Why is then only the sedimentation at the first time split counted as precep (for instance jn == 1 at line 464, but happens for all sedimentation, including the original bulkmicro and simpleice2 schemes, i think)

General limiting question. Is there a preference for using the mass (q_ci), or mass + tendency (q_ci + q_cip)? Using the m variables (n_clm etc) like line 462, vs the 0 variables, line 3000

[jchylik]

modbulkmicro_point.f90

* [ ]  l_sb_inuc_expl not implemented (line 465)

• [x] T, the warning is missing. Will be added after namelist reading

* [ ]  output of tendencies _dq_ci_dep_, _dq_hs_dep_. _dq_hg_dep_ is after the cor_deposit3 correction, line 870

• [ ] F, output of the deposition tendencies should be after correction.

[jchylik]

modbulkmicro_point.f90

• [ ] if not l_sb_lim_aggr, E_ab was unset? setting it to zero now. line 942

• [x] T, in the array version, E_ab was initialize with 0.

* [ ]  elseif statement should be an else. line 1367

• [x] T

* [ ]  if not _(D_cl.gt.D_c_a).and.(D_hg.gt.D_i0)_, can we turn off coll_gcg3? line 1474

• [x] T, in that case collection at given gridpoint does not happen.

* [ ]  if not _(D_cl.gt.D_c_a).and.(D_ci.gt.D_i0)_, can we turn off coll_ici3? line 1219

• [x] T, in that case collection at given gridpoint does not happen.

* [ ]  if not _(D_cl.gt.D_c_a).and.(D_hs.gt.D_i0)_, can we turn off coll_scs3? line 1323

• [x] T, in that case collection at given gridpoint does not happen.

* [ ]  evapmelting3: threshold for melting snow, should that be q_hs.gt.qicemin? line 2229

• [x] T, it should be qsnowmin

* [ ]  evapmelting3: threshold for melting graupel, should that be q_hg.gt.qicemin? line 2247

• [x] T, it should be qgrmin

* [ ]  accretion3: when not _l_sb_classic_, should accretion happen when _q_hr < q_hr_min_? line 2501

• [x] T Accretion in the old scheme happens when ((q_hr > q_hr_min) .and.(( q_cl> qcmin))

* [ ]  evap_rain3: is the use of unlimited x_hrf correct? line 2617

• [ ] F That is intentional. It was already limited by x_hrf = q_hr /(n_hr+eps0). When replaced by x_hr, it can lead in case of many big raindrops to removal of too many of them. (x_hr<x_hrf => phi/x_hr>phi/x_hrf )

* [ ]  satadj3: limit on _dn_cl_sa_ is this now correct? is used to include svp(in_cl) (to include tendencies from outside the scheme?). I now initialize n_clp from svp(in_cl). line 2718 and 2751

• [x] T: That makes sense. If n_clp is initialized with svp(in_cl), the svp here is not needed anymore.

* [ ]  sb_evmelt3: line 2997 and 3000, should the limit also include the tendency? _-n_cl/delt - n_clp_

• [x] T: the best would be to correct (-n_em/delt)-n_ep where n_ep is n_hsp, n_hgp, n_cip and n_em is svm(in_hs) etc.

* [ ]  integrals_bulk3: do we need to limit x_hr? line 370 and 389

• [x] F, we should keep limiting it because of calculations of heterogeneous freezing

modbulkmicro_column.f90

* [ ]  nucleation3: original code went out of bound (memory corruption) for d ssat / dk . line 130. Is this better?

• [x] F, it looks good now, but it would make sense to calculate just do k=2,k1-1 We do not want cloud nucleation at the first model layer (and at the top one none is expected).

* [ ]  line 360. Sedimentation threshold for rain

* [ ]  line 528. Sedimentation threshold for snow

* [ ]  line 726. Sedimentation threshold for ice

* [ ]  line 831. Sedimentation threshold for cloud water

• [ ] T, The correct limit should be smaller than qrmin (or q_hs_min etc.), but not yet tested what works best.

* [ ]  line 631, line 634. wrong variable (typo) be_hs => be_hg

• [x] T: that was a typo

* [ ]  lines 368 and 385, 412, Dvr_spl and x_hr are not calculated for different settings. Do we need to limit them, too?

• [x] T, That is a very good point, Dvr_spl should be calculated from xr_spl

[jchylik]

* [ ]  General sedimentation issue. The time-splitting scheme is to prevent stuff falling faster than 1 k-level. Why is then only the sedimentation at the first time split counted as precep (for instance _jn == 1_ at line 464, but happens for all sedimentation, including the original bulkmicro and simpleice2 schemes, i think)

A very good question. I kept it consistent with the warm microphysics, and never actually used this statistic. I can look at it later.

General limiting question. Is there a preference for using the mass (q_ci), or mass + tendency (q_ci + q_cip)? Using the m variables (n_clm etc) like line 462, vs the 0 variables, line 3000

It is generally preferable to limit the tendencies by variables m + update i.e. ( -(n_clm/delt+n_clp) )

[jiskattema]

Thanks for the comments. I've updated my code accordingly. I'll close this issue, and will use separate issues for the remaining questions (and new ones..)