DECKv1 Initialization and final tuning

Back to overview of v1 DECK simulations (low-res, water cycle)

Initialization of DECK piControl simulation

Due to time constraints, piControl year 1 (0001-01-01) initial conditions are taken from a pre-existing simulation rather than starting a new spin-up from scratch. The initial conditions are 0331-01-01 of 20171228.beta3rc13_1850.ne30_oECv3_ICG.edison.

Provenance of initial conditions:

• MPAS-O stand-alone initialization
• 10 years of G-case (20170814.FCTfix-del4.GMPAS-IAF.T62_oECv3.anvil)
  • Ocean and sea-ice restart as initial conditions for
• 50 years coupled FCT2 (0001-0051; 20170926.FCT2.A_WCYCL1850S.ne30_oECv3.anvil)
  • Year 51 ocean and sea-ice restart as initial conditions for
• 150 years of coupled beta3rc10 (0001-0151; 20171122.beta3rc10_1850.ne30_oECv3_ICG.edison)
  
  • Year 151 atmosphere, land, river, ocean and sea-ice restart as initial conditions for
• 100 years of coupled beta3rc11 (0151-0251; /wiki/spaces/SIM/pages/408453940)
  
  • Year 251 atmosphere, land, river, ocean and sea-ice restart as initial conditions for
• 80 years of coupled beta3rc13 (0251-0331; 20171228.beta3rc13_1850.ne30_oECv3_ICG.edison)
  
  • Year 331 atmosphere, land, river, ocean and sea-ice restart as initial conditions for
• DECK v1 piControl

Final tuning

At the time piControl needed to start, we were still in the process of fine tuning the control simulation to reduce its long term drift in surface air temperature as much as possible. 20171228.beta3rc13_1850.ne30_oECv3_ICG.edison was the configuration with the smallest drift at the time. However, due to its limited duration (80 years), we couldn't conclusively determine whether the drift was as small as it could be. As a result, we decided to start two piControl simulations with the goal of running both of them for 100 years and then selecting one:

• 20180129.DECKv1a_piControl.ne30_oEC.edison : identical tuning as in /wiki/spaces/SIM/pages/478412855 ,but with a slightly updated version of the code base.
• 20180129.DECKv1b_piControl.ne30_oEC.edison : same, except for a minor retuning of clubb_c14: 1.08 → 1.06. This is expected to only change net TOA fluxes by a few hundreds of a W/m2, but may help further reduce long-term drift (without hopefully changing its sign).

On 15 Feb 2018, both v1a and v1b piControl simulations completed 110 years. Analysis of the global surface temperature trend for years 11 to 110 revealed that v1b had the smallest drift of the two. Comparison of climatologies for years 51-100 between the two configurations showed almost no differences, as expected. As a result, v1b was selected as the final tuning for the Water Cycle low-res model.

Log

• 29 Jan 2018 : initial configuration and testing
• 30 Jan 2018 : simulation started
• 15 Feb 2018 : v1b is selected as the final tuning due to its reduced drift.

Edison jobs

20180129.DECKv1a_piControl.ne30_oEC.edison

20180129.DECKv1b_piControl.ne30_oEC.edison

Code tag/hash

2e145acfa7eaf89d2480c3967a150d4e15cc86e1 (from master)

Run scripts

run_acme.20180129.DECKv1a_piControl.ne30_oEC.edison.csh

run_acme.20180129.DECKv1b_piControl.ne30_oEC.edison.csh

Output files

Files on Edison scratch:

/global/cscratch1/sd/golaz/ACME_simulations/20180129.DECKv1{a,b}_piControl.ne30_oEC.edison

Sub-directories

archive/ : short-term archived files

run/ : raw model output files

post/ : post-processed files

E3SM Diags

AMWG

A-Prime for DECKv1a

• Time series (1-50), climatology (1-50)
• Time series (1-100), climatology (51-100)

A-Prime for DECKv1b

• Time series (1-50), climatology (1-50)
• Time series (1-100), climatology (51-100)