ICB 
---

ICB is a very idealized configuration used to test and debug the icb module.
The configuration is box with a shallow shelf (40m) on the east and west part of the domain with a deep central trough (> 100m).
ICB are generating using the test capability of the icb model along a E-W line (this can easily be tuned).

In this configuration, ONLY the ICB module is activated. It means that:
- SI3 is OFF
- NO atmospheric fluxes. See usr routine in MY_SRC.
- NO time variation of ocean properties. T, S, U, V, ssh (...) are intialized and never changed in time. This is achieved by simplifying the step.F90 routine in MY_SRC.

Resolution and Initial condition are:
- Homogeneous S (34)
- Spatially constant T     set to 2     over the first 2 levels and 1     below
- Spatially constant (U,V) set to (1,0) over the first 2 levels and (0,1) below
- e3 is set to 10m
- iceberg thicness is set to 100 in the class tested (10 level affected).
- iceberg density is set to 1026 as the ocean reference density => icb tikness = icb draft (easier to evaluate the output)

Results expected are:
- ln_M2016 = F => icb moving along the W direction only and pilling up on the W side of the domain.
- ln_M2016 = T => icb on shallow bank move at 45 degrees (NW) and 76 degrees (NNNNW) in the deeper through. 
- ln_M2016 = T and ln_icb_grd = T => as before but now icb cannot move onto the shallow bank on the W and migrate northward along the shelf break. Some get stuck at specific point because of the grounding law (they always get back to there intial position at every time step).

This set up allow us to easily test various option of the ICB code as:
- ln_M2016   (use properties in depth to compute dyn and thd)
- ln_icb_grd (use icb bottom level of the icb (tail) to ground the icb or not)

It can also help us to:
- reproduce bugs (interpolation issue for example)
- develop and test new functionalities (new grounding scheme for exemple)
- evaluate ICB performance alone

The side effect of the implementation is that:
- the run.stat (if produced) is meaningless as NEMO tracer and dynamics are not computed