Error with calculating o2_decomp_depth in bedrock

[ekluzek]

muszala < bugzillaMuszala > - 2013-06-11 06:02:41 -0600
Bugzilla Id: 1746
Bugzilla CC: erik, muszala, rfisher,

ERS_D.f10_f10.ICLM45BGCNoVS.yellowstone_intel.clm-rootlit

Ran in clm4_5_09. After the refactor in clm4_5_10, this test still runs but is not BFB with clm4_5_09. Consider clm4_5_09 truth.

As a sanity check, ERS_D.f10_f10.ICLM45BGCNoVS.frankfurt_intel.clm-rootlit is BFB between clm4_5_09 and clm4_5_10.

This looks like a problem specific to yellowstone & the intel compiler.

[ekluzek]

muszala < bugzillaMuszala > - 2013-06-12 13:05:28 -0600

found error with nag in

ch4Mod.F90, line 1003: Subscript 2 of POT_F_NIT_VR (value 2) is out of range (1:1).

fix from Jim:: change

if (.not. lake )

t0

if (.not. lake .and. j<=nlevdcomp_full ) then

fixes things with nag+frankfurt, intel+frankfurt and intel+yellowstone.

This will be added in an upcoming tag.

Consider any simulation with this compset suspect until the fix is in place.

Stef-

[ekluzek]

muszala < bugzillaMuszala > - 2013-06-13 09:49:25 -0600

put ch4Mod.F90 fix in clm4_5_12. Tested with:

frankfurt+nag
frankfrut+intel

yellowstone+intel

The fix does change CLM fields for this test only. Coupler and CLM history files remained BFB in all other tests. Please see the clm4_5_12 ChangeLog for details.

[ekluzek]

muszala < bugzillaMuszala > - 2013-06-13 10:25:48 -0600

SPM- reopened. proper fix is needs to go in a future tag.

Charlie brought me into this thread. As I mentioned to him, I think what was going on is, the methane model is supposed to work without the vertically resolved BGC (indeed, it was originally developed in that mode), but I didn't anticipate that "NITRIF_DENITRIF" would be used without vertically resolved BGC when I wrote that loop. If this is an unsupported configuration intended for machine testing only, then Jim's fix below is fine; otherwise I would tweak it to get the correct behavior, as Jim's fix will increase anoxia in the top soil layer and reduce methane oxidation more than intended. Better to just ignore the oxygen demand by nitrifiers when the vertically resolved soil BGC is not used, and replace the logic with something like:

"if (.not. lake .and. nlevdcomp >= nlevsoi) then"
or
"if (.not. lake .and. nlevdcomp_full >= nlevgrnd) then"

--Zack

[ekluzek]

Erik Kluzek < erik > - 2014-05-22 23:21:54 -0600

*** Bug 1672 has been marked as a duplicate of this bug. ***

[ekluzek]

Erik Kluzek < erik > - 2014-05-23 00:17:55 -0600

Here's what those changes look like in cesm1_2_2

Index: models/lnd/clm/src/clm4_5/biogeochem/ch4Mod.F90
===================================================================
--- models/lnd/clm/src/clm4_5/biogeochem/ch4Mod.F90	(revision 60566)
+++ models/lnd/clm/src/clm4_5/biogeochem/ch4Mod.F90	(working copy)
@@ -674,7 +674,7 @@
    use ch4varcon,        only : q10ch4base, q10ch4, rootlitfrac, cnscalefactor, mino2lim, & 
                                 f_ch4, lake_decomp_fact, usephfact, anoxicmicrosites, ch4rmcnlim
    use clm_varctl,       only : anoxia
-   use clm_varpar, only : nlevdecomp
+   use clm_varpar, only : nlevdecomp, nlevdecomp_full, nlevgrnd
 #ifdef CENTURY_DECOMP
    use CNDecompCascadeMod_CENTURY, only : nlev_soildecomp_standard
 #else
@@ -999,7 +999,7 @@
 
          ! Add oxygen demand for nitrification
 #if (defined NITRIF_DENITRIF)
-         if (.not. lake) then
+         if (.not. lake .and. nlevdecomp_full >= nlevgrnd) then
             o2_decomp_depth(c,j) = o2_decomp_depth(c,j) + pot_f_nit_vr(c,j) * 2.0_r8/14.0_r8
                                                         ! g N/m^3/s           mol O2 / g N
          end if

[ekluzek]

Erik Kluzek < erik > - 2014-05-23 13:46:50 -0600

Zack's fix was put into: cesm1_2_x_n12_clm4_5_10

[ekluzek]

Bill Sacks < sacks > - 2016-05-16 14:13:44 -0600

Zack's suggested fix never made it to the trunk; it should.

[ekluzek]

From looking at the code, it looks like an equivalent fix was put into place. So closing this...