search

ESCOMP / CTSM

Community Terrestrial Systems Model (includes the Community Land Model of CESM)
http://www.cesm.ucar.edu/models/cesm2.0/land/
Other
307 stars 310 forks source link

Use of C13/C14 timeseries for non-transient control cases gives incorrect input when run long enough #592

Open ekluzek opened 5 years ago

ekluzek commented 5 years ago

Brief summary of bug

Some 1850 control simulations with Carbon isotopes turned on the C13/C14 time-series files. Any of these simulations that ran for more than 1850 would have used the wrong values for the carbon isotopes.

General bug information

CTSM version you are using: any version

Does this bug cause significantly incorrect results in the model's science? Yes (for C13/C14)

Configurations affected:

1850 or 2000 compsets that have CLM4.5/CLM5.0 and isotopes with the time-series on

non-transient simulations that use the C13/C14 time-series as follows...

use_c13_timeseries = .true.
use_c14_bombspike = .true.

will be using constant 1850 conditions until you run for more than 1850 years, when it will start running through the time-series.

Details of bug

Note the reason for this is due to #182, where the hardcoded mechanism for reading in isotopes was fragile and allowed little control. Resolving that issue can resolve this problem as well as make it able to work with 2000 control, which we can NOT do at this time. For 2000 conditions isotopes will be running with 1850 conditions, and if you turn the time-series on as above, you'll have to run for 2000 years before you are running with 2000 conditions, and then it will fly by until you reach the end of the file.

Note Esther Brady pointed this limitation out on Nov/21st, and we realized it was a bug at that time.

mingjieshi commented 5 years ago

Hi Erik,

I checked the time-series of delta C14 respirations (e.g., delta_C14ER = ((c14ER/ER)/1e-12 - 1) * 1e3) over Arctic (the historical runs are at /gpfs/fs1/p/cgd/tss/people/oleson/CLM_LAND_ONLY_RELEASE/CLM5/clm50_r270_1deg_GSWP3V1_iso_newpopd_hist/lnd/proc/tseries/month_1/), and found earlier peaks (during 1935-1945) than I expected (1960s). Do you think if this bug was the reason for the early peak? Thanks!

Mingjie

billsacks commented 2 years ago

Blocked: depends on #182 (in principle we could do this without #182 but #182 makes this significantly easier).