Abstract. The ensemble adjustment Kalman filter (EAKF) is used to estimate the erodibility fraction parameter field in a coupled meteorology and dust aerosol model (Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS)) over the Sahara desert. Erodibility is often employed as the key parameter to map dust source. It is used along with surface winds (or surface wind stress) to calculate dust emissions. Using the Saharan desert as a test bed, a perfect model Observation System Simulation Experiments (OSSEs) with 40 ensemble members, and observations of aerosol optical depth (AOD), the EAKF is shown to recover correct values of erodibility at about 80% of the points in the domain. It is found that dust advected from upstream grid points acts as noise and complicates erodibility estimation. It is also found that the rate of convergence is significantly impacted by the structure of the initial distribution of erodibility estimates; isotropic initial distributions exhibit slow convergence, while initial distributions with geographically localized structure converge more quickly. Experiments using observations of Deep Blue AOD retrievals from the MODIS satellite sensor result in erodibility estimates that are considerably lower than the values used operationally. Verification shows that the use of the tuned erodibility field results in better predictions of AOD over the west Sahara and the Arabian Peninsula.

The Ensemble Adjustment Kalman Filter (EAKF) is used to estimate the erodibility fraction parameter field in a coupled meteorology and dust aerosol model (Coupled Ocean Atmosphere Mesoscale Prediction System-COAMPS) over the Sahara desert. Erodibility is often employed as the key parameter to map dust source. It is used along with surface winds (or surface wind stress) to calculate dust emissions. Using the Saharan desert as a test bed, a perfect model Observation System Simulation Experiments (OSSEs) with 40 ensemble members, and observations of aerosol optical depth (AOD), the EAKF is shown to recover correct values of erodibility at about 80% of the points in the domain. It is found that dust advected from upstream grid points acts as noise and complicates erodibility estimation. It is also found that the rate of convergence is significantly impacted by the structure of the initial distribution of erodibility estimates; isotropic initial distributions exhibit slow convergence while initial distributions with geographically localized structure converge more quickly. Experiments using observations of Deep Blue AOD retrievals from the MODIS satellite sensor result in erodibility estimates that are considerably lower than the values used operationally. Verification shows that the use of the tuned erodibility field results in better predictions of AOD over the Western Sahara and Arabia

Abstract. The implications of state dependent, finite time error growth has been studied using singular values in a chaotic 2-dimensional map. Earlier studies have demonstrated the superiority of the singular values over the Lyapunov number in representing error growth over finite time scales, since they take state dependency into account. In this work, linearized error growth as given by singular values under operational constraints like non-isotropic initial uncertainty and model error is considered. It is demonstrated that the relevant singular values in the case of non-isotropic initial uncertainty are quite different from the isotropic case. The effect of model inadequacy on error growth is delineated.

Abstract. In this study, we present the development of a new coupled weather and carbon monoxide (CO) data assimilation system based on the Environment and Climate Change Canada (ECCC) operational ensemble Kalman filter (EnKF). The estimated meteorological state is augmented to include CO. Variable localization is used to prevent the direct update of meteorology by the observations of the constituents and vice versa. Physical localization is used to damp spurious analysis increments far from a given observation. Perturbed surface flux fields are used to account for the uncertainty in CO due to errors in the surface fluxes. The system is demonstrated for the estimation of three-dimensional CO states using simulated observations from a variety of networks. First, a hypothetically dense, uniformly distributed observation network is used to demonstrate that the system is working. More realistic observation networks, based on surface hourly observations, and space-based observations provide a demonstration of the complementarity of the different networks and further confirm the reasonable behavior of the coupled assimilation system. Having demonstrated the ability to estimate CO distributions, this system will be extended to estimate surface fluxes in the future.