Global near-real-time CO2 forecast
MACC-II has added a new daily product to its large portfolio of global forecast products. Global atmospheric CO2 concentrations are now being forecasted with a five-day lead time at very high resolution (~16 km in the horizontal, 91 vertical levels) every 3 hours. The forecast uses the infrastructure of the European Centre for Medium Range Weather Forecasts (ECMWF) Integrated Forecasting System (IFS). It is run every day and it is initialized two days behind real-time in order to be able to use near-real-time biomass burning emissions estimated from satellite observations. The atmospheric CO2 is cycled from one forecast to the next, whereas the meteorology is re-initialized with ECMWF operational analyses at the beginning of each forecast. The model uses CO2 fluxes that are prescribed from inventories and from off-line statistical and physical models. The terrestrial biogenic fluxes are modelled online within the IFS, as part of the CTESSEL Carbon module. The anthropogenic fluxes are based on annual mean fluxes from the EDGAR database v4.2 in 2008, extrapolated using a rescaling factor to account for the annual trend. The monthly mean Takahashi et al. (2009) CO2 ocean flux climatology is also used. Finally, daily mean biomass burning fluxes are obtained from the GFASv1.0 database. These are kept fixed throughout the 5-day forecast.
Plots of the CO2 forecast every 6-hours for the past 30 days are available from here. A near-real-time evaluation of the forecast is provided online using observations from the ICOS network.
NRT verification using ICOS observations
The impact of using high horizontal resolution is also shown by comparing the high resolution forecast with a low resolution one (~80 km and 60 vertical levels) with the same configuration.
Hurricane Sandy simulation
The CO2 forecast fields can also be accessed and requested via the MACC-II catalogue. These can be used as prior concentration fields to assimilate measurements from space or in situ; they can also be used as boundary conditions for regional modeling, as well as providing prior information for satellite retrievals and planning the sampling in the current observing system of CO2.