Sea ice changes

[aekiss]

Time series of sea ice extent and volume. Maps of concentration and thickness change. *Comparison with obs.

Previous issue using old simulations is https://github.com/pedrocol/basal_mom5-collaborative-project/issues/64.

[aekiss]

Plots below are from this version ofnotebooks/seaice/seaice.ipynb updated in PR https://github.com/pedrocol/basal_mom5-collaborative-project/pull/80.

Plot descriptions are mostly copied from https://github.com/pedrocol/basal_mom5-collaborative-project/issues/64, with differences in italic.

[aekiss]

Here are timeseries of the sea ice volume over the full length of the experiments, integrated over the full Antarctic or just the shelf (onshore of 1000m isobath). They stabilise after about 3 years, so in the later plots I use 1903-1909 (inclusive) for climatologies.

There are two pairs:

1. without Gade: Control and _BasalNoGade_NoIcb are nearly identical (blue overlaps black, very hard to see)_
2. with Gade: BasalGade_NoIcb and BasalGade_NoIcb_Brine have more ice than without Gade

The difference between these pairs is more pronounced on the shelf. _Brine rejection at depth (BasalGade_NoIcb_Brine) brings the ice volume closer to Control on the shelf, roughly halving the impact of Gade._

[aekiss]

These are the 1903-1909 monthly climatologies of sea ice extent (i.e. area with concentration over 15%), compared to observations over the RYF forcing period (1 May 1990 - 30 April 1991; NB: the obs data is not a monthly climatology). Differences between the 4 experiments are much smaller than the bias relative to obs. Model sea ice is too low in the Feb minimum, rises too soon to a peak that's close to obs but too early (Aug instead of Sept), then falls off initially too gradually and then too steeply. The Gade experiments seem to have the smallest bias, but are not significantly improved over the others.

[aekiss]

_Here are the 1903-1909 mean thickness differences from Control. Thickness in BasalGade_NoIcb (left column) is mainly increased on the West Antarctic shelves, but also near the coast in some East Antarctic regions._ As seen in the volume timeseries, BasalNoGade_NoIcb (middle) is nearly identical to Control, but with slight thinning in the Amundsen-Bellingshausen seas. _BasalGade_NoIcb_Brine (right column) is similar to BasalGade_NoIcb (left), but weaker._

Annual mean patterns (top row) are similar to those in Feb (middle row) and Sept (bottom). Thinning is mostly in winter (Sept), whereas thickening seems less dependent on season. May need to look at more months to confirm this.

[aekiss]

The 1903-1909 sea ice concentration differences seem more seasonal than the thickness differences. _In contrast to thickness, the SIC change in BasalGade_NoIcb_Brine (right column) does not resemble BasalGade_NoIcb (left), except for the pattern of SIC increase in summer (middle row)_ in the Weddell, eastern Ross and Prydz shelves. _Winter changes (bottom) are mainly in the MIZ, and of opposite sign for BasalGade_NoIcb_Brine (right) and BasalGade_NoIcb (left) north of the Amundsen and Ross seas._ _Unlike for thickness, BasalNoGade_NoIcb (middle) is not significantly closer to Control, except for a general absence of places where SIC increases in summer (middle row)._

[aekiss]

1903-1909 monthly climatological sea ice concentration biases (relative to observations over the RYF forcing period 1 May 1990 - 30 April 1991; NB: the obs data is not a monthly climatology) are very similar in the 4 experiments. February has a pronounced low bias, especially in the Weddell Sea, which is nearly identical for the four models. September biases are smaller and mostly in the ice edge. _BasalGade_NoIcb_Brine seems very slightly better than the others in this season._

[aekiss]

I've updated these plots to use standard expt names, climatology period, line styles - see notebooks/seaice/seaice.ipynb

[aekiss]

I'd hoped to plot congel and frazil but they weren't saved for 01deg_jra55v13_ryf9091_DSW.