Coupled model runs start from September initial dump

[penguian]

resolution_fixed | by mrd599@nci.org.au

---

All the coupled model suites have used the same initial dump as the AMIP runs,

AINITIAL = /g/data/access/TIDS/OCEAN/hadgem3/initial/atmos/N96L85/ab357a.da19810901_00

However coupled models runs start on Jan 1 rather than Sep 1 so there's a mismatch here. The reconfiguration changes the dump file date to 0001-01-01 but soil temperature etc will still be September values.

---

Issue migrated from trac:342 at 2024-01-31 18:32:07 +1100

[penguian]

@martin.dix@anu.edu.au commented

---

Would expect that this will particularly affect the deep soil temperatures.

In u-al623 the level 4 soil temperature in Jan year 1 is much colder than the time mean in the SH and warmer in the NH. Differences are much smaller in year 2.

The global mean RMS difference wrt the climatology reaches equilibrium in about 10 years (some of this may be temperature drift unrelated to the date error).

RMS difference over the region 60S to 60N reaches equilibrium in just a few years.

[penguian]

@martin.dix@anu.edu.au changed _comment0 which not transferred by tractive

[penguian]

@martin.dix@anu.edu.au uploaded file tdiff1.png (56.1 KiB)

[penguian]

@martin.dix@anu.edu.au uploaded file tdiff2.png (51.7 KiB)

[penguian]

@martin.dix@anu.edu.au _uploaded file rms_series.png (59.8 KiB)_

[penguian]

@martin.dix@anu.edu.au _uploaded file rms_series2.png (48.8 KiB)_

[penguian]

@martin.dix@anu.edu.au commented

---

The install_cold task in the coupled model suite resets the dump file date with

        script = echo -e "$BYEAR 1 1\n" | ~access/apps/pythonlib/umfile_utils/change_dump_date.py $ROSE_DATA/$RUNID.astart

The UM configuration has

[namelist:headers]
i_override_date_time=1
new_date_time=$BYEAR,5*0

which means that the reconfiguration resets the year number. Without the extra explicit override the model would have picked up the mismatch in months. E.g.

?  Error code: 10
?  Error from routine: INITTIME
?  Error message:
?        Mismatch between model_basis_time read from namelist and validity time 
read
?        from dump fixed header.
?
?        model_basis_time    =      1      1      1      0      0      0
?        fixhd validity time =      1      9      1      0      0      0

See also #344.

[penguian]

@martin.dix@anu.edu.au changed _comment0 which not transferred by tractive

[penguian]

@martin.dix@anu.edu.au commented

---

Starting from a GA7/7.1 dump file with stochastic physics variables requires the reconfiguration fix described in comment:3:#344.

[penguian]

@martin.dix@anu.edu.au changed status from new to closed

[penguian]

@martin.dix@anu.edu.au set resolution to fixed

[penguian]

@martin.dix@anu.edu.au commented

---

All pre-industrial suites are descended from u-aq959 which starts from a Jan 1st dump from an AMIP run,

  AINITIAL = /projects/access/access-cm2/input_b/um_astart/aj458a.da19890101_00