Statistical Optimization of an Atmospheric GCM coupled to a Reduced Gravity Model of the Upper Tropical Ocean

One of the challenges to studying the effects of climate change on tropical climate variability is the high degree of sensitivity that predictions from models of the coupled atmosphere-ocean system have to detailed features of the thermocline. Arguably, all coupled GCM models of the atmosphere-ocean system provide poor simulations of tropical climate variability. In order to extend our ability to predict how changes in climate affect tropical variability we consider how one may go about tuning a new experimental framework that includes a global atmospheric GCM that has been coupled to a reduced gravity model of the upper ocean between 30N and 30N with a slab ocean polewards of these latitudes. The primary advantage of this model is the explicit control the scientist has over processes affecting thermocline dynamics without sacrificing on the ability to use state-of-the-art models of the atmosphere. Wang, Jackson, and Chang (publication in preparation) have recently completed a study using an efficient stochastic sampling algorithm akin to a Monte Carlo Markov Chain to quantify and enumerate all model configurations (sets of uncertain model parameters) that would permit a reduced order model of the upper tropical ocean to reproduce the observed variability of the Nino3 index over the past 150 years. The present study extends this exercise so that solutions are sought that 1) reproduce the observed sea surface temperature variance, skewness, kurtosis, and power at interannual and decadal frequencies for all points within the tropical ocean (30S to 30N) and 2) are optimized for an ocean system that has been coupled to the CCM3.10 atmospheric GCM at T42 resolution. We attempt this optimization through a repetition of steps that begin with evaluating fields that are necessary for coupling atmosphere and ocean dynamics, calculating optimized ocean model parameters from off-line (uncoupled) experiments, and recalculating the coupling fields from a revised coupled model experiment. One of the goals of this study is to determine whether or not this sequence of steps would converge to a reasonable solution. We also plan to highlight how aspects of tropical climate variability are related to processes governing thermocline dynamics and modeling assumptions.