The Diffusion Kernel Filter Applied to Lagrangian Data Assimilation

Krause, Paul; Restrepo, Juan M.
December 2009
Monthly Weather Review;Dec2009, Vol. 137 Issue 12, p4386
Academic Journal
The diffusion kernel filter is a sequential particle-method approach to data assimilation of time series data and evolutionary models. The method is applicable to nonlinear/non-Gaussian problems. Within branches of prediction it parameterizes small fluctuations of Brownian-driven paths about deterministic paths. Its implementation is relatively straightforward, provided a tangent linear model is available. A by-product of the parameterization is a bound on the infinity norm of the covariance matrix of such fluctuations (divided by the grid model dimension). As such it can be used to define a notion of “prediction” itself. It can also be used to assess the short time sensitivity of the deterministic history to Brownian noise or Gaussian initial perturbations. In pure oceanic Lagrangian data assimilation, the dynamics and the statistics are nonlinear and non-Gaussian, respectively. Both of these characteristics challenge conventional methods, such as the extended Kalman filter and the popular ensemble Kalman filter. The diffusion kernel filter is proposed as an alternative and is evaluated here on a problem that is often used as a test bed for Lagrangian data assimilation: it consists of tracking point vortices and passive drifters, using a dynamical model and data, both of which have known error statistics. It is found that the diffusion kernel filter captures the first few moments of the random dynamics, with a computational cost that is competitive with a particle filter estimation strategy. The authors also introduce a clustered version of the diffusion kernel filter (cDKF), which is shown to be significantly more efficient with regard to computational cost, at the expense of a slight degradation in the description of the statistics of the dynamical history. Upon parallelizing branches of prediction, cDKF can be computationally competitive with EKF.


Related Articles

  • A METHOD OF NON-LINEAR TIME SERIES PREDICTION. BIELIŃSKA, EWA // Systems Science;2008, Vol. 34 Issue 2, p11 

    The paper concerns time series modelling and prediction. Time series dynamics is described with the use of a hybrid linear-bilinear model. Then, a prediction algorithm that minimizes a variance of a prediction error is derived based on the hybrid linear-bilinear model, and the method of...

  • Linear and non-linear filtering in mathematical finance: a review. Date, P.; Ponomareva, K. // IMA Journal of Management Mathematics;Jul2011, Vol. 22 Issue 3, p195 

    This paper presents a review of time series filtering and its applications in mathematical finance. A summary of results of recent empirical studies with market data are presented for yield curve modelling and stochastic volatility modelling. The paper also outlines different approaches to...

  • A New Hybrid Methodology for Nonlinear Time Series Forecasting. Khashei, Mehdi; Bijari, Mehdi // Modelling & Simulation in Engineering;2011, Vol. 2011, p1 

    Artificial neural networks (ANNs) are flexible computing frameworks and universal approximators that can be applied to a wide range of forecasting problems with a high degree of accuracy. However, using ANNs to model linear problems have yielded mixed results, and hence; it is not wise to apply...

  • Weakly nonlinear shock propagation in slowly varying one-dimensional flows. Prasad, Dilip // Physics of Fluids;Mar2006, Vol. 18 Issue 3, p036101 

    A theoretical model for the linear and nonlinear evolution of acoustic perturbations in nonuniform one-dimensional isentropic flows is developed. On the assumption that the length scale of the mean flow greatly exceeds that of the imposed disturbances, it is shown that the propagation of linear...

  • ESTIMATION LINEAR MODEL USING BLOCK GENERALIZED INVERSE OF A MATRIX. Jasinska, Elzbieta; Preweda, Edward // Proceedings of the International Multidisciplinary Scientific Ge;2013, p167 

    The work shows the principle of generalized linear model, point estimation, which can be used as a basis for determining the status of movements and deformations of engineering objects. The structural model can be put on any boundary conditions, for example, to ensure the continuity of the...

  • Application of a variational iteration method to linear and nonlinear viscoelastic models with fractional derivatives. Drăgănescu, G. E. // Journal of Mathematical Physics;Aug2006, Vol. 47 Issue 8, p082902 

    Two nonlinear anelastic models with fractional derivatives, describing the properties of a series of materials as polymers, and polycrystalline materials are presented in this paper. These models are studied analytically, using a variational iteration method. The paper clarifies the different...

  • Nonlinear Generalization of Singular Vectors: Behavior in a Baroclinic Unstable Flow. Rivière, Olivier; Lapeyre, Guillaume; Talagrand, Olivier // Journal of the Atmospheric Sciences;Jun2008, Vol. 65 Issue 6, p1896 

    Singular vector (SV) analysis has proved to be helpful in understanding the linear instability properties of various types of flows. SVs are the perturbations with the largest amplification rate over a given time interval when linearizing the equations of a model along a particular solution....

  • The Strong Second-Order Sufficient Condition and Constraint Nondegeneracy in Nonlinear Semidefinite Programming and Their Implications. Defeng Sun // Mathematics of Operations Research;Nov2006, Vol. 31 Issue 4, p761 

    For a locally optimal solution to the nonlinear semidefinite programming problem, under Robinson's constraint qualification, the following conditions are proved to be equivalent: the strong second-order sufficient condition and constraint noedegeneracy: the nonsingularity of Clarke's Jacobian of...

  • Fixed-size Least Squares Support Vector Machines: A Large Scale Application in Electrical Load Forecasting. Espinoza, Marcelo; Suykens, Johan; Moor, Bart // Computational Management Science;Mar2006, Vol. 3 Issue 2, p113 

    Based on the Nyström approximation and the primal-dual formulation of the least squares support vector machines, it becomes possible to apply a nonlinear model to a large scale regression problem. This is done by using a sparse approximation of the nonlinear mapping induced by the kernel...


Read the Article


Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics