Open geeseproblems opened 10 months ago
What version of MARBL are you using? I can confirm that the latest commit on development
has non-zero values for that field:
$ git checkout development
Switched to branch 'development'
Your branch is up to date with 'origin/development'.
$ cd tests/regression_tests/call_compute_subroutines/
$ ./call_compute_subroutines.py
[snip compilation details]
(run_exe): Running following command:
(run_exe): /Users/mlevy/NO_BACKUP/codes/MARBL/tests/driver_exe/marbl.exe -n test.nml -u cgs -s ../../input_files/settings/marbl_with_o2_consumption_scalef.settings
Beginning call_compute_subroutines test...
* MARBL output will be written to history_1inst.nc
Instance 1 has 5 columns, beginning with 1
-------------
Timer summary
-------------
There are 4 timers being returned
----
MARBL Init: 0.044 seconds
MARBL surface_flux_compute: 0.000 seconds
MARBL interior_tendency_compute: 0.001 seconds
MARBL carbonate chemistry: 0.000 seconds
$ ncdump -v graze_sp_poc_zint_100m history_1inst.nc | tail -n 3
graze_sp_poc_zint_100m = 0.00987925742038663, 0.00301135262980637, 0,
0.00232910295184793, 0.00583966155888692 ;
}
On stable
I made the following mods:
$ git diff .
diff --git a/src/marbl_interior_tendency_mod.F90 b/src/marbl_interior_tendency_mod.F90
index e6017ef..de36629 100644
--- a/src/marbl_interior_tendency_mod.F90
+++ b/src/marbl_interior_tendency_mod.F90
@@ -2978,6 +2978,25 @@ subroutine compute_particulate_terms(k, domain, &
POC%remin(k) = c0
POC%sflux_out(k) = c0
POC%hflux_out(k) = c0
+ P_CaCO3%remin(k) = c0
+ P_CaCO3%sflux_out(k) = c0
+ P_CaCO3%hflux_out(k) = c0
+ P_CaCO3_ALT_CO2%remin(k) = c0
+ P_CaCO3_ALT_CO2%sflux_out(k) = c0
+ P_CaCO3_ALT_CO2%hflux_out(k) = c0
+ P_SiO2%remin(k) = c0
+ P_SiO2%sflux_out(k) = c0
+ P_SiO2%hflux_out(k) = c0
+ dust%remin(k) = c0
+ dust%sflux_out(k) = c0
+ dust%hflux_out(k) = c0
+ P_iron%remin(k) = c0
+ P_iron%sflux_out(k) = c0
+ P_iron%hflux_out(k) = c0
+ POP%remin(k) = c0
+ POP%sflux_out(k) = c0
+ POP%hflux_out(k) = c0
+ PON_remin = c0
endif
because our compilers are uninitialized memory that used to slip by (this was fixed on development
via #372) and I'm still seeing non-zero graze_sp_poc_zint_100m
:
$ ncdump -v graze_sp_poc_zint_100m history_1inst.nc | tail -n 3
graze_sp_poc_zint_100m = 0.00987925742038663, 0.00301135262980637, 0,
0.00232910295184793, 0.00583966155888692 ;
}
graze_sp_poc_zint_100m
and the equivalent variables for other PFTs are all zeros - it seems they are not being calculated in the diagnostics mod.