The main program, ice.F, contains the time-stepping loop. The model flow at the highest level is shown below.
ICEMODEL-+-initialize variables
|
do time-stepping loop: +-------<-------+
| | | +---<----+ |
| +-receive info from coupler | v | |
| | |fsalt_hist=0| |
| +-vertical thermodynamics calculations | | | |
| | | v | |
| +-send ice state and fluxes to coupler | | | |
| | | | | |
| +-thickness redistribution thermodynamics | | | |
| | fsalt=0 | ^ ^
| +-dynamics calculations | | | |
| | v | | |
| +-write std out to log file | | | |
| | | | | |
| +-write avg quantities to history file | +--->----+ |
| | +------->-------+
end time-stepping loop
|
+-exit_coupler
A complete calling tree is shown in Appendix A. The ice model receives information from the coupler at the beginning of the time loop. The vertical ice thermodynamics are calculated, and the ice state and fluxes are returned to the coupler at mid-timestep immediately after they are computed. The information is sent to the coupler at this point in the time-stepping loop for load balancing. This allows the atmosphere to begin calculations instead of waiting until the end of the ice model timestep. The thermodynamics calculated after returning data to the coupler are mostly related to the ice thickness distribution and include transport in thickness space, lateral growth and melting, and freeboard adjustment.
As shown in the model flow in Section 2.1, fluxes are sent to the coupler in the middle of the timestep, and the diagnostics are written out at the end of the timestep. Most fluxes, like the atmospheric latent heat, are initialized at the beginning of the timestep and calculated only in the thermodynamics. This is the value that is sent to the coupler and then added to the average being calculated for the history file.
A problem arises for fluxes that are modified before and after the fluxes are sent to the coupler, that is, in both the thermodynamics and dynamics calculations. These fluxes are fsalt, fresh, fhnet, and fswthru. If these values are initialized at the start of the time step, the values computed in the dynamics during the previous time step would be written over. The fluxes sent to the coupler would not represent quantities summed over an entire timestep. This is why these fluxes are initialized immediately after the call to the coupler.
If these fluxes (fsalt, fresh, fhnet, fswthru) were written to the history file, the fluxes computed at the beginning of the step (before the call to the coupler) would be neglected. The variables with _hist appended to the name are initialized before the thermodynamics calculation and measure the same thing as their coupler counterparts but over different time intervals. The _hist variables contain information calculated from the start to the end of a timestep and are written to the history file.
To calculate the global sums for the diagnostics, there is a value of the ice concentration aice_init that is saved from the beginning of the timestep.