Antarctic Bottom Water in CMIP5 models: characteristics, formation, evolution
The first part of my thesis is an extended version of Heuzé et al. (2013):
- assessment of bottom temperature, salinity and hence density biases in CMIP5 models over 1986 – 2005 (historical run);
- study of shelf processes in CMIP5 models, in particular whether these models form and export dense shelf water;
- brief study of other ways to form Antarctic Bottom Water (further studied in a later chapter)
- comparison of the different methods to define the mixed layer.
It then contains Heuzé et al. (2015a) which studies the changes in global bottom water as projected by CMIP5 models by the RCP4.5 and 8.5 climate change scenarios.
I briefly studied in more details two puzzling CMIP5 models with respect to their Southern Ocean deep water formation:
- CCSM4, which features an original “pipe” parameterisation to artificially bring dense shelf water into the deep ocean. That way, dense shelf water retains its properties, instead of being too mixed like the other CMIP5 models. Unfortunately, other biases (in particular in the sea ice) prevent us from advocating for a widespread use of this parameterisation.
- INMCM4, which has neither open ocean deep convection nor dense shelf water export. We are still unsure as to whether it even forms any Antarctic Bottom Water.
Finally, I tried to understand what triggers open ocean deep convection in climate models:
- first by comparing (with no success) models from the same family, HadGEM2 and HiGEM,
- then by performing sensitivity experiments on the ocean component of HadGEM3 as described in Heuzé et al. (2015b)