SIC budget analysis in the CICE model

[Yongwu]

I have one model experiment is performed from 1980 to 2020 without any sea ice data assimilation. Two types of output are saved, the monthly mean and daily mean. The monthly output includes sea ice concentration and other diagnostic variables (e.g., sea ice area tendency due to thermodynamical (daidtt) and dynamical process(daidtd)). However, the daily output only includes sea ice concentration, with no daidtt or daidtd data saved.

My questions:

• Can I utilize daidtt and daidtd from the monthly output to calculate sea ice concentration changes between two days or two months? For example, in April 1980, is the cumulative effect of (daidtt + daidtd) over 30 days the same as the sea ice concentration difference between April 1st and April 30th ?
• How many processes are included in the daidtt and daidtd? In other words, what is the formula of daidtt and daidtd embedded in CICE?

I try to find relevant information, but unfortunately, I have not been successful thus far.

Best wishes,
Yongwu

 

[dbailey]

The terms daidtt and dadtd contain all of the thermodynamic (step_therm1, step_therm2) and dynamic (ridging, advection) processes respectively. You can see where these are computed in ice_step_mod.F90. The monthly mean is not really the same as April 30th - April 1st. I think your best bet is to rerun the experiment and add these to the daily stream.

 

[mafolb]

Hello,
I am also struggling to understand the area tendency terms daidtt and daidtd, even looking at ice_step_mod.F90 and the CICE documentation which present the thickness distribution equation.

• Does a negative daidtd correspond to a sea ice area loss from dynamic processes (advection, ridging) or is it actually sea ice area gain (negative daidtd corresponding to convergence)? From the simulations I am using, daidtd seems to remain negative throughout the year north of the Canadian Arctic Archipelago, where sea ice convergence should take place in winter from the main currents in the Arctic and sea ice consolidation in the Archipelago.
• From monthly outputs, can I compute the mean melt season integrated (say May to October) sea ice area changes melt from daidtt*tarea*number_days_per_month + daidtd*tarea*number_days_per_month?
• The tendency terms seem to allow aice exceeding 1 integrated over a month i.e aice + daidtt*number_days_per_month + daidtd*number_days_per_month > 1. Should I cap daidtt and daidtd when integrating (sum) in time? I am working with monthly values, but I should recover mean diagnostic aice from daidtt and daidtd... there is something I am missing.

Thank you very much,
Madeleine

 

[Yongwu]

Hi Madeleine,

Thanks for your question. From my side, I compare monthly daidtd & daidtd with two SNAPSHOT of the model output. Fortunately, it works.

An example to illustrate what I try to do:
I have two snapshot outputs of the model state(let's say 2008-04-01-00000 and 2008-05-01-00000) and daidtd/daidtt for the 2008-04 monthly mean. What I found is the following relationship: (daidtd + daidtt) * numbers_in_April_day = Model_snapshot_at_2008-05-01-00000 minus Model_snapshot_at_2008-04-01-00000.

So, for your question:
1) I'm not sure, but I do think so, i.e., when sea ice convergence, the sea ice concentration/area will decrease.
2) From May to October, daidtt*tarea*number_days_per_month + daidtd*tarea*number_days_per_month equals the difference of the snapshot November-01-00000 minus May-01-00000. I'm not sure whether this value is exactly what you want or not.
3) After understanding what the daidtd and daidtt mean, I think you will not be surprised that the aice + daidtt*number_days_per_month + daidtd*number_days_per_month > 1. This is because aice is the monthly mean while the sum of daidtt/daidtd*number_days_per_month is the difference of snapshot. They have different views of the model state.

To be honest, I don’t know how to leverage the daidtt/daidtd for climate research. For example, if I found a sea ice extent drop from the 2008-04 monthly mean to 2008-05 monthly mean, till now I don’t know how to use the daidtt/daidtd to understand which process led to this drop because the daidtd/daidtt is the difference between two snapshots while the widely-used SIE in particular months like 2008-04 2008-05 is the monthly mean.

Hope this could help you!
Best,
Yongwu

 

[dbailey]

Sorry, I missed this question. We are currently involved in the CICE Consortium workshop. I will try to get to this next week.

 

[mafolb]

Hi Yongwu and David,

Thank you both for your answers! I think I have understood; correct me if I am wrong:
CICE solves a thickness distribution equation in time including the dynamic tendency term, the thermodynamic tendency term and the ridging distribution term. However, when this equation is transferred in the change of SIA, there is only thermodynamic and dynamic tendency terms. The dynamic area tendency includes gain and loss through ridging. A negative daidtd - the ridging contributions (+ dardg1dt, - dardg2dt) will give something more intuitive with positive (convergence) and negative (divergence) area tendencies corresponding to main sea ice patterns in the Arctic.

Understanding this, integrating with monthly mean values of daidtd and daidtt over the melt season should give the sea ice loss before and after the melt season.

Best,
Madeleine

 

[Yongwu]

Sorry, I missed this question. We are currently involved in the CICE Consortium workshop. I will try to get to this next week.

Hello, have you come up with a solution for this question? That is, how to use the daidtd and daidtt with the monthly mean sea ice concentration variable? and the question with the meaning of dardg1dt, - dardg2dt. Many thanks.