Ocean-atmosphere coupling in supermodel exploits opposing wind effects

Atmosphere and ocean flows are characterized by interacting turbulent motions from the planetary down to the micrometer scale, but can’t all be explicitly modeled due to computational constraints. Climate modeling requires some approximations in order to represent effects of unresolved scales, and each differ in their approach.

Combining outputs from different models often improves results. The so-called supermodeling approach does this with interactive combinations that connect and synchronize models dynamically. By balancing opposing dynamical interactions, supermodeling can create new, more accurate solutions and can outperform predictions from the linear combinations of separate outputs, despite exhibiting the same kind of errors. But supermodeling is considerably more difficult to accomplish.

For the Intertropical Convergence Zone (ITCZ), an important climate feature central to the well-known El Nino Southern Oscillation climate variation, a report in Chaos detangles the role of ocean-atmosphere coupling effects in supermodeling. Acquiring an accurate description of these interactions is a long standing challenge.

The authors coupled two different atmospheric models to a single ocean model, revealing nonlinear interactions in the region that provided new insights into the role of supermodel wind effects. These winds affect ocean dynamics, like the upwelling of cold water at the equator, and significantly impacted the ITCZ simulation. While most studies of the same region focus on the equatorial wind stress, these results demonstrated a crucial impact from wind stress at latitudes off the equator balancing the influence of equatorial wind stress.

Machine learning was key in combining the wind effects of the different convection schemes used in models constituting the supermodel. Repeated simulations with differing weighting schemes trained the supermodel to account for the complexities of the dynamically interacting elements.

For even more complex models of regions with chaotic and unpredictable atmospheres, Mao-Lin Shen, one of the work’s authors, looks forward to “connecting 3-D dynamic states of multiple atmosphere models and to coupling multiple ocean models in order to synchronize models of the full globe.”

Source: “Role of atmosphere-ocean interactions in supermodeling the Tropical Pacific climate,” by Mao-Lin Shen, Noel Keenlyside, Bhuwan C. Bhatt, and Gregory S. Duane, Chaos: An Interdisciplinary Journal of Nonlinear Science (2017). The article can be accessed at https://doi.org/10.1063/1.4990713 .