A regional carbon data assimilation system and its preliminary evaluation in East Asia

Abstract

Abstract. In order to optimize surface CO 2 fluxes at grid scales, a regional surface CO 2 flux inversion system (Carbon Flux Inversion system and Community Multi-scale Air Quality, CFI-CMAQ) has been developed by applying the ensemble Kalman filter (EnKF) to constrain the CO 2 concentrations and applying the ensemble Kalman smoother (EnKS) to optimize the surface CO 2 fluxes. The smoothing operator is associated with the atmospheric transport model to constitute a persistence dynamical model to forecast the surface CO 2 flux scaling factors. In this implementation, the "signalto-noise" problem can be avoided; plus, any useful observed information achieved by the current assimilation cycle can be transferred into the next assimilation cycle. Thus, the surface CO 2 fluxes can be optimized as a whole at the grid scale in CFI-CMAQ. The performance of CFI-CMAQ was quantitatively evaluated through a set of Observing System Simulation Experiments (OSSEs) by assimilating CO 2 retrievals from GOSAT (Greenhouse Gases Observing Satellite). The results showed that the CO 2 concentration assimilation using EnKF could constrain the CO 2 concentration effectively, illustrating that the simultaneous assimilation of CO 2 concentrations can provide convincing CO 2 initial analysis fields for CO 2 flux inversion. In addition, the CO 2 flux optimization using EnKS demonstrated that CFI-CMAQ could, in general, reproduce true fluxes at grid scales with acceptable bias. Two further sets of numerical experiments were conducted to investigate the sensitivities of the inflation factor of scaling factors and the smoother window. The results showed that the ability of CFI-CMAQ to optimize CO 2 fluxes greatly relied on the choice of the inflation factor. However, the smoother window had a slight influence on the optimized results. CFI-CMAQ performed very well even with a short lag-window (e.g. 3 days).

Improving PM<sub>2. 5</sub> forecast over China by the joint adjustment of initial conditions and source emissions with an ensemble Kalman filter

“…It is noted although the method is very similar to that used by Peters et al (2007) and Peng et al (2015) for CO 2 emission inversion, it is still of novelty for applications in aerosol anthropogenic emissions. In Peters et al (2007), λ p i,t were all 1.…”

“…In Peng et al (2015), the CO 2 DA system worked comparatively well when the ensemble spread of λ a i,t ranged from 0.05 to 1.25 for β = 60, 70, 75, 80. The assimilated CO 2 fluxes deviated markedly from the "true" CO 2 fluxes when the ensemble spread of λ a i,t were too small for β = 10, 50 or when the ensemble spread of λ a i,t were too large for β = 100.…”

“…As no suitable dynamic model was available to forecast the emission scaling factors, a persistence forecasting operator served as the forecast model for the scaling factors, similar to the method used by Peng et al (2015) for CO 2 emission inversion. Figure 2a shows the flowchart for the persistence forecasting operator M SF .…”

“…Besides, the framework is more advanced compared to our previous work. In Peng et al (2015), in order to generate λ p i,t , a set of ensemble forecasts were performed from time t to t + 1 to produce the CO 2 concentration fields, forced by the prescribed net CO 2 surface fluxes with the previous assimilated concentration fields as initial conditions. That means that the ensemble forecast was performed twice in that DA system, which time consuming.…”

A fundamental theorem of Earth’s surface modelling

“…In addition to CarbonTracker, a regional surface CO 2 flux inversion system (Carbon Flux Inversion system and Community Multi-scale Air Quality, CFI-CMAQ) was developed by applying the ensemble Kalman filter (EnKF) to constrain the CO 2 concentrations and applying the ensemble Kalman smoother (EnKS) to optimize the surface CO 2 fluxes (Peng et al 2015). A framework, termed TanTracker, was developed for assimilating observations of atmospheric CO 2 concentrations on the basis of an ensemble four-dimensional variation data assimilation method (Tian et al 2014), which is based on proper orthogonal decomposition.…”