Time travelling ice does the following:
- first order conservatively map the current ice fractions $I^o_j$ to the atmosphere grid $I^a_j$
- divide the atm->ice fluxes by this ice fractions (on atm grid) before first order conservatively mapping to the ice grid
- scale these fluxes by the ice fractions (on the ocean grid) $I^o_j$
This works well except for points which $I^a_j = 0$ and $f_j \neq 0$. At these points we currently take the CM2 approach and do not scale these values before conservatively re-mapping. These fluxes get mapped to ice free cells ($I^a_j = 0$ implies that all ocean cells overlapping this atmosphere cell are ice free). Therefore these fluxes get "lost".
The fluxes from these cells should be mapped directly to ocean instead to preserve energy and mass.
Notation:
- superscript
a and o refer to quantities on the atm and ocn (/ice) grid repectively (ocn and ice share the same grid)
- subscripts refer to grid points
-
$f$ is an atm->ice flux (grid box mean, not ice area mean)
-
$I$ is the ice fraction
-
$A$ is the cell area
Time travelling ice does the following:
This works well except for points which$I^a_j = 0$ and $f_j \neq 0$ . At these points we currently take the CM2 approach and do not scale these values before conservatively re-mapping. These fluxes get mapped to ice free cells ($I^a_j = 0$ implies that all ocean cells overlapping this atmosphere cell are ice free). Therefore these fluxes get "lost".
The fluxes from these cells should be mapped directly to ocean instead to preserve energy and mass.
Notation:
aandorefer to quantities on the atm and ocn (/ice) grid repectively (ocn and ice share the same grid)