Articles | Volume 16, issue 16
Research article
18 Aug 2016
Research article |  | 18 Aug 2016

A biogenic CO2 flux adjustment scheme for the mitigation of large-scale biases in global atmospheric CO2 analyses and forecasts

Anna Agustí-Panareda, Sébastien Massart, Frédéric Chevallier, Gianpaolo Balsamo, Souhail Boussetta, Emanuel Dutra, and Anton Beljaars

Abstract. Forecasting atmospheric CO2 daily at the global scale with a good accuracy like it is done for the weather is a challenging task. However, it is also one of the key areas of development to bridge the gaps between weather, air quality and climate models. The challenge stems from the fact that atmospheric CO2 is largely controlled by the CO2 fluxes at the surface, which are difficult to constrain with observations. In particular, the biogenic fluxes simulated by land surface models show skill in detecting synoptic and regional-scale disturbances up to sub-seasonal time-scales, but they are subject to large seasonal and annual budget errors at global scale, usually requiring a posteriori adjustment. This paper presents a scheme to diagnose and mitigate model errors associated with biogenic fluxes within an atmospheric CO2 forecasting system. The scheme is an adaptive scaling procedure referred to as a biogenic flux adjustment scheme (BFAS), and it can be applied automatically in real time throughout the forecast. The BFAS method generally improves the continental budget of CO2 fluxes in the model by combining information from three sources: (1) retrospective fluxes estimated by a global flux inversion system, (2) land-use information, (3) simulated fluxes from the model. The method is shown to produce enhanced skill in the daily CO2 10-day forecasts without requiring continuous manual intervention. Therefore, it is particularly suitable for near-real-time CO2 analysis and forecasting systems.

Short summary
This paper presents a method to adjust the sinks and sources of CO2 associated with land ecosystems within a global atmospheric CO2 forecasting system in order to reduce the errors in the forecast. This is done by combining information on (1) retrospective fluxes estimated by a global flux inversion system, (2) land-use information, and (3) simulated fluxes from the model. Because the method is simple and flexible, it can easily run in real time as part of a forecasting system.
Final-revised paper