HR: 17:05h
AN: NG44A-04 INVITED [Abstracts]
TI: Model Assessment With Lagrangian Metrics and Data
AU: * Kirwan, A D
EM: adk@udel.edu
AF: University of Delaware, Robinson Hall, Newark, DE 19716, United States
AU: Lipphardt, B L
EM: brucel@udel.edu
AF: University of Delaware, Robinson Hall, Newark, DE 19716, United States
AU: Poje, A C
EM: poje@wiener.math.csi.cuny.edu
AF: City University of New York, Staten Island, New York, NY 19716, United States
AU: Kantha, L
EM: kantha@colorado.edu
AF: University of Colorado, Dept of Aerospace Engineering, Boulder, CO 19716, United States
AU: Zweng, M
EM: mzweng@udel.edu
AF: University of Delaware, Robinson Hall, Newark, DE 19716, United States
AB:
As geophysical predictive models typically are Eulerian, it seems natural to evaluate their performance with
Eulerian metrics and observations. For example, meteorological predictions are assessed by how well they
predict precipitation or temperature at specific locations while oceanographic models often are evaluated by
comparison of predicted currents with current meter moorings and predicted water mass properties with CTD
casts. In the oceanographic case, assessment often is problematic since most such models are exercised in the
forward mode. A true test of model skill requires data assimilation since it is then possible to compare model
results directly with independent observations of specific ‘events'. Predictive models, judged successful by
Eulerian metrics, often fail to perform as well when required to predict Lagrangian properties such as the paths of
hurricanes, the motion of ocean eddies, the dispersion of drifting sensor arrays, and the movement of
contaminants in the environment. Model assessment with Lagrangian data and metrics, which are inherently
nonlinear, is an emergent issue in oceanography. Here, two types of metrics, along with appropriate data are
used to assess a data-assimilating model of ocean currents. The first type, the prediction of individual
trajectories, generally show poor performance when compared with observations. On the other hand, methods
adapted from dynamical systems theory show remarkable ability to account for the outbreak of chlorophyll plumes
and the dispersion of drifting sensor arrays. We also compare model predictions of eddy formation, breakup and
the movement of cyclones around large anticyclones with conventional Eulerian observations.
DE: 3238 Prediction (3245, 4263)
DE: 4445 Nonlinear differential equations
DE: 4475 Scaling: spatial and temporal (1872, 3270, 4277)
DE: 4499 General or miscellaneous
DE: 4534 Hydrodynamic modeling
SC: Nonlinear Geophysics [NG]
MN: 2007 Joint Assembly