zuibeidemei@gmail_com
New Member
Dear all:
I have a question about the subcycling in CAM5 eul dynamics core.
But I cannot find any related person who should be responsible for this.
Can anyony suggest me how to find support.
The question is posted below:
In normal condition, eul_nsplit =1. So if dtime=1200, then ztodt=2400 (except the first step), then both the dynamics and physics
package use 2400s to do a time-stepping for acheiving vars at n3. This is because the semi-implicit dynamics require 3 time level.
When eul_nsplit>1, say 10. Then ztodt = dtime=1200 (used for one large step physics). As the following code,
the dynamics have different way. The dt_dyn0 is always 120. If it's the first large step, then dt_dyn=240.
Then the actual first small dynamics step use dt=120, and the following small step in the first large step use dt=240, which is double dt.
This is the same way as that in eul_nsplit=1.
But when it goes further (from the second large step), the dt_dyn=dt_dyn0=120 for always.
This suggests that each dynpkg call will use vars at n3m2 and n3m1 to produce n3.
But the time-interval between n3m2 and n3 should be 240, not 120.
So I have some confusion about this.
I have a question about the subcycling in CAM5 eul dynamics core.
But I cannot find any related person who should be responsible for this.
Can anyony suggest me how to find support.
The question is posted below:
In normal condition, eul_nsplit =1. So if dtime=1200, then ztodt=2400 (except the first step), then both the dynamics and physics
package use 2400s to do a time-stepping for acheiving vars at n3. This is because the semi-implicit dynamics require 3 time level.
When eul_nsplit>1, say 10. Then ztodt = dtime=1200 (used for one large step physics). As the following code,
the dynamics have different way. The dt_dyn0 is always 120. If it's the first large step, then dt_dyn=240.
Then the actual first small dynamics step use dt=120, and the following small step in the first large step use dt=240, which is double dt.
This is the same way as that in eul_nsplit=1.
But when it goes further (from the second large step), the dt_dyn=dt_dyn0=120 for always.
This suggests that each dynpkg call will use vars at n3m2 and n3m1 to produce n3.
But the time-interval between n3m2 and n3 should be 240, not 120.
So I have some confusion about this.