Posts Tagged ‘Ensemble Kalman filter’

Up the ante on bioeconomic submodels of marine food webs

September 20, 2016

My new paper is available for free download for 50 days (until November 9, 2016). The paper is published in the journal Ecological Economics and discusses modeling of marine food webs such that economic analysis is viable. At the core of our approach lie the ensemble Kalman filter, something I have used earlier. In this new application, we go further in reducing model parameter dimensionality and move beyond the filtering routine to estimate certain structure parameters. We also apply a data transformation that deal with previously overlooked endogeneity in stock level data. We use all this to estimate a model of the largest pelagic fish stocks in the Norwegian Sea. The abstract:

eeWhile economists have discussed ecosystem-based fisheries management and similar concepts, little attention has been devoted to purposeful modeling of food webs. Models of ecosystems or food webs that make economic analysis viable should capture as much as possible of system structure and dynamics while balancing biological and ecological detail against dimensionality and model complexity. Relevant models need strong, empirical content, but data availability may inhibit modeling efforts. Models are bound to be nonlinear, and model and observational uncertainty should be included. To deal with these issues and to improve modeling of ecosystems or food webs for use in ecosystem-based fisheries management analysis, we suggest the data assimilation method ensemble Kalman filtering. To illustrate the method, we model the dynamics of the main, pelagic species in the Norwegian Sea. In order to reduce parameter dimensionality, the species are modeled to rely on a common carrying capacity. We also take further methodological steps to deal with a still high number of parameters. Our best model captures much of the observed dynamics in the fish stocks while the estimated model error is moderate.

The paper is part of the EINSAM project.

The Ensemble Kalman Filter for Multidimensional Bioeconomic Models

September 22, 2015

Natural Resource ModelingIn the recent issue of Natural Resource Modeling, I have an article together with my old boss. The idea is to apply the ensemble Kalman filter to fit multidimensional foodweb models to data for use in bioeconomic analysis. The abstract:

To integrate economic considerations into management decisions in ecosystem frameworks, we need to build models that capture observed system dynamics and incorporate existing knowledge of ecosystems, while at the same time accommodating economic analysis. The main constraint for models to serve in economic analysis is dimensionality. In addition, to apply in long-term management analysis, models should be stable in terms of adjustments to new observations. We use the ensemble Kalman filter to fit relatively simple models to ecosystem or foodweb data and estimate parameters that are stable over the observed variability in the data. The filter also provides a lower bound on the noise terms that a stochastic analysis requires. In this paper, we apply the filter to model the main interactions in the Barents Sea ecosystem. In a comparison, our method outperforms a regression-based approach.

The Ensemble Kalman Filter

October 27, 2010

The standard Kalman filter and even the Extended Kalman filter (for nonlinear problems) proved inadequate. I’ve now placed my hope in what’s known as the Ensemble Kalman Filter:

Another sequential data assimilation method which has received a lot of attention is named the Ensemble Kalman Filter (EnKF). The method was originally proposed as a stochastic or Monte Carlo alternative to the deterministic [Extended Kalman filter] by Evensen (1994a).  The EnKF was designed to resolve the two major problems related to the use of the [Extended Kalman filter] with nonlinear dynamics in large state spaces, i.e. the use of an approximate closure scheme and the huge computational requirements associated with the storage and forward integration of the error covariance matrix.

The EnKF gained popularity because of its simple conceptual formulation and relative ease of implementation, e.g. it requires no derivation of a tangent linear operator or adjoint equations and no integrations backward in time. Furthermore, the computational requirements are affordable and comparable to other popular sophisticated assimilation methods […].*

* Excerpt from Geir Evensen’s Data Assimilation: The Ensemble Kalman Filter, 2007, p. 38.