Climate modelling, large-scale ocean circulation, climate uncertainty
I am an ocean modeller with a particular interest in the role of the ocean in climate variability and climate change.
I am an active contributor to the Joint Marine Modelling Project (JMMP), which is developing, jointly between NERC and the Met Office, the ocean component of the next-generation Earth System Model. The first deliverable of this project was the GO5.0 ocean configuration (Megann et al., 2014), with a ¼° global resolution, allowing the model to represent ocean eddies. The most recent deliverable is GO6.0 (Storkey et al, 2018), which forms the ocean component of the GC3.1 coupled climate model and the UKESM1 Earth System Model, both aimed at the IPCC Sixth Assessment Report (AR6) and the associated Coupled Model Intercomparison Project (CMIP6). GO6 includes a representation of icebergs, and a grid that is extended southwards to allow it to model the Antarctic ice sheets and their interactions with the ocean. Development of the next generation, GO8, is ongoing, and this will include several model improvement including interactive marine ice shelves as well as new features to reduce numerical mixing. I am leading WP3.2, which is concerned with the model physics improvements.
A current active research project is to diagnose the numerical mixing (mixing arising from truncations in the model advection scheme) in the NEMO ocean model. This mixing is in addition to that imposed by the mixing scheme of the model, the latter aiming to reproduce the mixing effect of real ocean processes such as breaking internal waves and shear instabilities. My results (Megann, 2018) indicate that numerical diapycnal mixing (mixing of water of different densities) can be several times larger than the explicit physical mixing. This has potential implications for the performance of the model in long climate projections, in particular on its ability to represent the uptake and storage of heat associated with anthropogenic climate change.
I lead the RENUMERATE project, funded under the CMEMS Service Evolution II programme. The aims of this are: (1) to add tidal forcing to the GO8 1/4° global model configuration; (2) to add the z~ flexible vertical coordinate, aimed at reducing numerical mixing from internal waves and tides; and (3) to evaluate the joint effects of these two additions on numerical mixing, watermass changes, large-scale ocean circulation and sea ice. This was the first implementation of either tidal forcing or z~ to a global 1/4° NEMO, and successfully demonstrated that z~ indeed has the intended effect in reducing spurious mixing in a tidal simulation.
I lead WP2.3 of the North Atlantic Climate System Integrated Study (ACSIS), which delivered integrations of the GO6 1/4° global NEMO ocean configurations from 1958 to the present day with three different forcing datasets. This showed a direct and unambiguous connection between recent changes in surface buoyancy fluxes in the Labrador and Irminger Seas and decadal changes in the Atlantic meridional overturning circulation (AMOC).
I led the modelling component of OSCAR, a large interdisciplinary project investigating the effect of geothermal heat input on ocean circulation. This was focused on the nearly enclosed Panama Basin in the eastern tropical Pacific, which is a region with a high level of hydrothermal activity, but also illuminated the influence of geothermal heating on global ocean circulation. OSCAR involved an observational campaign involving hydrographers and geophysicists, as well as a high-resolution regional implementation of the HYCOM ocean model to simulate the circulation in the basin.
I created the CHIME coupled climate model (Megann et al, 2010), using the HYCOM hybrid-coordinate isopycnic ocean coupled to the atmosphere component of the HadCM3 climate model. Isopycnic-coordinate ocean models, using layers of constant density, have the potential for much reduced numerical mixing compared with the default model type, which is based on constant depth levels. This project showed that the isopycnic ocean in CHIME had a much more realistic representation of climatically important water masses such as North Atlantic Deep Water, Subantarctic Mode Water and Antarctic Intermediate Water than the HadCM3 model, identical to CHIME apart from its depth-coordinate ocean component. CHIME also simulated the overflow of Arctic water over the sills between Greenland and Scotland into the North Atlantic without the unphysical convection and excessive entrainment of the overflow waters seen in HadCM3 and similar models.
I have a strong Interest in climate uncertainty, particularly that associated with the ocean. I was involved in a joint project between NOC and the University of Exeter on developing parameter ensembles of the NEMO ocean model at different grid resolutions, and the creation of Bayesian emulators. This technique generates optimum ranges for the model parameters, but in addition allows a relatively large ensemble of a computationally efficient low-resolution model to supplement a small ensemble of a more realistic (but more expensive) high-resolution model to optimise the parameters of the latter.