Ocean Atmosphere General Circulation Models (OAGCM) exhibit significant model biases, e.g., the tropical bias in the Intertropical Convergence Zone cold tongue complex. A significant part of such biases is due to uncertainties in the model parameters. Thus far, however, parameter tuning in OAGCMs remains rudimentary. This collaborative project will develop a novel strategy for systematic parameter optimization in OAGCMs, using an ensemble based data assimilation strategy, focusing on the outstanding tropical bias problem. Building on an existing coupled data assimilation system, an adaptive coupled ensemble Kalman Filter system (AcEnKF) for a simultaneous estimation of model parameter and state, will be developed. Different assimilation techniques to search for an effective method of parameter optimization in OAGCMs will be studied, as also coupled dynamics relevant to parameter optimization in the tropical climate system for both the climatology and climate variability. The work is expected to give the first comprehensive evaluation of the ensemble based strategy for parameter optimization in OAGCMs. The study will also shed light on the mechanism of the coupled dynamics of the tropical climate system and its sensitivity to model parameters. In spite of potential difficulties, this pilot study represents a significant step forward in the improvement of OAGCMs. This activity has significant broader impacts: it will lay a foundation for a systematic improvement of OAGCMs and, more generally, future coupled earth system models.
In spite of decades of effort, coupled ocean-atmosphere general circulation models (OAGCM) still suffer from significant model biases. A significant part of these biases can potentially, however, be reduced by optimizing model parameters. So far, however, parameter tuning in OAGCMs remains rudimentary. Here, we have explored a novel strategy for systematic parameter optimization in an OAGCM. We explored the parameter optimization in two lines of research in coupled climate models using the Ensemble Kalman Filter (EnKF). First, in a fully coupled ocean-atmosphere general circulation model, we demonstrate a first successful parameter estimation using an EnKF scheme. Second, in an intermediate climate model, the reduction of model bias associated with the biased model physics is studied using parameter optimization within a biased "twin" experiment framework. It is shown that the error due to the bias of the model climatology and the consequent climate forecast can be reduced significantly by parameter optimization. Together, these results suggest that parameter optimization may help us improving model climate and seasonal to decadal predictions in coupled general circulation models that include imperfect physical schemes and model parameters. Our work has presented the first successful parameter estimation in OAGCMs. In spite of potential difficulties, our pilot study represents a significant step forward in the improvement of OAGCMs. Our ensemble based scheme, we believe, will provide the most promising and practical approach towards a systematic improvement of an OAGCM. Therefore, our work will lay a foundation for a systematic improvement of OAGCMs and, more generally, coupled earth system GCMs, in the future.
