Use of Genetic Algorithms for Ocean Model Parameter Optimisation

When working with Earth system models, a considerable challenge that arises is the need to establish the set ofparameter values that ensure the optimal model performance in terms of how they reflect real-world observed data. Giventhat each additional parameter under investigation increases the dimensional space of the problem by one, simple brute-forcesensitivity tests can quickly become too computationally strenuous. In addition, the complexity of the model and interactionsbetween parameters mean that testing them on an individual basis has the potential to miss key information. As such, this5work argues the need of the development of a tool that can give an estimation of parameters. Specifically it proposes the useof a Biased Random Key Genetic Algorithm (BRKGA). This method is tested using the one dimensional configuration ofPISCES-v2, the biogeochemical component of NEMO, a global ocean model. A test case of particulate organic carbon in theNorth Atlantic down to 1000 m depth is examined, using observed data obtained from autonomous biogeochemical Argo floats.In this case, two sets of tests are run, one where each of the model outputs are compared to the model outputs with default10settings, and another where they are compared with 3 sets of observed data from their respective regions, which is followed bya cross reference of the results. The results of these analyses provide evidence that this approach is robust and consistent, andalso that it provides indication of the sensitivity of parameters on variables of interest. Given the deviation of the optimal set ofparameters from the default, further analyses using observed data in other locations are recommended to establish the validityof the results obtained.