Milija Zupanski, Colorado State University

Coupled data assimilation with Maximum Likelihood Ensemble Filter (MLEF)


Data assimilation with coupled modeling systems is an important application of data assimilation. One of the major challenges of coupled data assimilation is adequate representation of the cross-component correlations. In this presentation we will discuss the relevance of coupled forecast error covariance and its significance for extracting the maximum information from observations, followed by the details of the approach taken in MLEF. We will also show recent applications of MLEF with (i) land surface-atmosphere and (ii) aerosol-chemistry-atmosphere coupled systems, using the Weather Research and Forecasting (WRF) model as the atmospheric component.

The assimilated observations include: atmospheric observations, aerosol optical depth (AOD) observations, and total column so2, no2, and o3 observations. We also examine the information content of assimilated observations.

The results suggest that using coupled ensemble-based forecast error covariance improves the utility of observation information, implying that coupled data assimilation system is more efficient than the sum of individual components. Error covariance localization considerations associated with applications to coupled systems will also be discussed.

File available here