Seminar- Quantifying CO2 flux variability using geostatistical ensemble filters: Current challenges and future potential

12/13/2011 - 3:30pm
ML- Main Seminar Room
81

Quantifying CO2 flux variability using geostatistical ensemble filters: Current challenges and future potential

Abhishek Chatterjee, University of Michigan

The ensemble approach has gained a lot of popularity within the carbon science community due to its simple conceptual formulation and relative ease of implementation. Hitherto most applications of the ensemble filter for CO2 flux inversions have revolved around estimating surface carbon fluxes at low resolutions without the use of an explicit dynamical model. As a result several gaps remain in our understanding of the behavior of the ensemble system for large scale CO2 inversions. These include among others, issues related to the ensemble size, assumptions related to covariance localization and inflation, and the representation of error covariances and their temporal propagation.

In this presentation, the impact of these numerical approximations and filter properties will be examined from the perspective of a geostatistical ensemble square root filter. Both covariance inflation and localization will be shown to be ‘necessary but not sufficient’ techniques for obtaining reliable flux estimates and uncertainties. Results indicate that the more critical parameters driving the CO2 ensemble system are the ensemble size, and in the absence of an underlying dynamical model, the spatial and temporal density of the measurement network used with the filter. Given the increasing availability of diverse data streams of CO2 and advanced earth system models, future prospects for transitioning from an inversion to an assimilation framework will also be discussed.