Abstract
We present a parametric Kalman filter data assimilation system using GOSAT methane observations within the hemispheric CMAQ model. The assimilation system produces forecasts and analyses of concentrations and explicitly computes its evolving error variance while remaining computationally competitive with other data assimilation schemes such as 4-dimensional variational (4D-Var) and ensemble Kalman filter (EnKF). The error variance in this system is advected using the native advection scheme of the CMAQ model and updated at each analysis while the error correlations are kept fixed. We discuss extensions to the CMAQ model to include methane transport and emissions (both anthropogenic and natural) and perform a bias correction for the GOSAT observations. The results using synthetic observations show that the analysis error and analysis increments follow the advective flow while conserving the information content (i.e., total variance). We also demonstrate that the vertical error correlation contributes to the inference of variables down to the surface. In a companion paper, we use this assimilation system to obtain optimal assimilation of GOSAT observations.
Highlights
Methane is the second most important greenhouse gas (GHG) after CO2, contributing to about 16% of the anthropogenic radiative forcing of all types of GHGs [1,2,3]
While most past studies have focused on the global inversion of methane sources at low spatial resolution, inversions over regional domains and at higher resolutions are important for constraining anthropogenic sources, such as fugitive and high-emitting sources from industry [19,20,21,22]
It is sufficient to define a correlation function in R3 × R3, so that any subspace of R3 define a correlation on that subspace. This property was used in this study to define underlying homogeneous isotropic correlation functions, which are mapped onto the polar stereographic grid of H-Community Multiscale Air Quality (CMAQ), which has a uniform grid spacing on the projected plane
Summary
Methane is the second most important greenhouse gas (GHG) after CO2 , contributing to about 16% of the anthropogenic radiative forcing of all types of GHGs [1,2,3]. Estimating the concentrations (i.e., in our case, the model state), whether for the purpose of constraining the initial conditions or boundary conditions or the whole domain in space and time, is a typical problem of data assimilation. 4D-Var chemical data assimilation methods (i.e., to estimate the state, not the emissions) was conducted using a stratospheric chemistry transport model [33,34] Another approach designed for the assimilation of long-lived chemical species is based on simplified Kalman filtering. In terms of algorithm computational efficiency, the applicability of the parametric Kalman filter (or PvKF) to large state-space is made from the correspondence between the continuous and discrete representations of the different operators, between spatial correlation functions and the corresponding correlation matrices This idea was originally developed for the Optimal Interpolation (OI) method.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
