CISL Seminar- Efficient Ensemble State-Parameters Estimation Techniques: Application to Ocean Ecosystem Models

12/10/2015 - 10:00am

Mohamad El Gharamti
Nansen Environmental and Remote Sensing Center (NERSC), Bergen, Norway

Given the recent strong international focus on developing efficient data assimilation frameworks for biological models, I will present in this comparative study the application of newly developed state-parameters estimation tools to two different ocean ecosystem models.

Standard joint state-parameters augmentation technique, using the ensemble Kalman filter (EnKF), often introduces significant inconsistency during the update especially for strongly nonlinear models. This is usually the case for ocean ecosystem models at the time of the spring bloom. Separating the probability density functions of the state and the parameters using the so-called Dual EnKF is often utilized as a better estimation strategy. The dual (heuristic) filter is computationally more expensive than the Joint EnKF but is expected to perform more accurately. Using a similar separation procedure, I propose a new EnKF algorithm in which I apply a one-step-ahead smoothing to the state. The new state-parameters estimation scheme is derived in a consistent Bayesian filtering framework and results in separate update steps for the state and the parameters.

The performance of the new smoothing-based schemes will be tested against the standard EnKF in one-dimensional configurations of the: (1) GOTM-NORWECOM and (2) Norwegian Earth System (NorESM) models. For constraining the models, I make use of nutrient in-situ profiles, satellite surface chlorophyll and partial CO2 measurements from three different stations in the Atlantic and Pacific oceans. Estimates of various biological parameters will be evaluated. Further, I will analyze the performance of the filters in terms of complexity and accuracy of the state and parameters’ estimates.

Date: December 10, 2015

Time: 10:00am

Location: Foothills Laboratory - Main Auditorium