Drifter modeling and error assessment in wind driven currents
MetadataShow full item record
As hydrodynamic models are used to predict the transport of contaminants, it follows that model validation should involve assessments of the model’s ability to predict observed Lagrangian transport pathways. Existing validation methods, however, often only involve comparisons of Eulerian field and model data, from which Lagrangian transport pathways are not always discernible. This research explores the use of drifters and particle tracking in assessing a hydrodynamic model’s ability to predict Lagrangian transport. The significant advances resulting from this research are the development of the Semi-Lagrangian model for predicting drifter transport, and the development of the Circle Assessment method for comparing sets of drifter paths. The advantages of the these advances were demonstrated while assessing the ability of the Estuary and Lake Computer Model (ELCOM) to reproduce drifter paths observed in Marmion Marine Park, Western Australia. The analyses indicated the ELCOM model was successful at reproducing the larger-scale features of the observed drifter movement to the degree that predicting movement was computationally achievable given the input data available. Through the Circle Assessment analyses of Semi-Lagrangian drifter results, possible ELCOM model deficiencies were also identified. The Semi-Lagrangian drifter model predicts drifter movement by determining the forces acting on the drifter by the surrounding fluids. The model incorporates the influence of winds and inertia on drifter motion. Existing drifter models assumed drifters moved in perfect accord with the surrounding currents or with velocities offset from the current velocities by some small fraction of the wind velocity. Such simplistic models are adequate for predicting larger-scale characteristics of waterbody circulation, but are too imprecise to accurately reproduce drifter paths over the shorter time an length scales. The Circle Assessment method for analyzing drifter data quantifies the model’s ability to reproduce the field drifter motion over both short and long time and length scales. The method also provides diagnostic information regarding model performance, which may suggest avenues for model improvement by changing the hydrodynamic model algorithms or setup. Existing assessment methods are only applicable over larger time and length scales, are qualitative, and do not provide diagnostic information regarding model behavior.