A Fuzzy-Based Framework for Assessing Uncertainty in Drift Prediction Using Observed Currents and Winds
This paper proposes a fuzzy number—based framework for quantifying and propagating uncertainties through a model for the trajectories of objects drifting at the ocean surface. Various sources of uncertainty that should be considered are discussed. This model is used to explore the effect of paramete...
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
Frontiers Media S.A.
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/a20a40a7eda44e02b24d43f13871bfd8 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | This paper proposes a fuzzy number—based framework for quantifying and propagating uncertainties through a model for the trajectories of objects drifting at the ocean surface. Various sources of uncertainty that should be considered are discussed. This model is used to explore the effect of parameterizing direct wind drag on the drifting object based on its geometry, and using measured winds to parameterize shear and rotational dynamics in the ocean surface currents along with wave-driven circulation and near-surface wind shear. Parameterizations are formulated in a deterministic manner that avoids the commonly required specification of empirical leeway coefficients. Observations of ocean currents and winds at Ocean Station Papa in the northeast Pacific are used to force the trajectory model in order to focus on uncertainties arising from physical processes, rather than uncertainties introduced by the use of atmospheric and hydrodynamic models. Computed trajectories are compared against observed trajectories from five different types of surface drifters, and optimal combinations of forcing parameterizations are identified for each type of drifter. The model performance is assessed using a novel skill metric that combines traditional assessment of trajectory accuracy with penalties for overestimation of uncertainty. Comparison to the more commonly used leeway method shows similar performance, without requiring the specification of empirical coefficients. When using optimal parameterizations, the model is shown to correctly identify the area in which drifters are expected to be found for the duration of a seven day simulation. |
|---|