Skip to Content

Combining observations and models to understand fluxes and trends

Combining observations and models to understand fluxes and trends


Observing networks of greenhouse gases and air pollutants consist of a sparse network of long-term in-situ surface measurements with high precision and accuracy or ground based remote sensing such as TCCON and NDACC, global coverage of satellite measurements, with larger errors and a weak vertical sensitivity, and occasional vertical sampling onboard aircraft. Earth system models integrate surface fluxes, atmospheric processes, such as transport and chemistry, to provide a 4-dimensional field of those quantities. However, these models are plagued by large uncertainties. Data assimilation and inverse modeling are now commonly used and they are powerful tools to synergistically combine observations and models to provide the best estimates of the atmospheric abundance and fluxes. However, the set of observations cannot provide enough constraints because of the large degrees of freedom in the earth system models. Thus, residuals errors in the posterior fields and/or fluxes can still be large. We will present two examples where we try to remediate and quantify those remaining errors, by using ensembles of runs, and by comparing model results with non-assimilated observations.

First, we will examine a global reanalysis of carbon monoxide (CO) that is based on a joint assimilation of conventional meteorological observations and Measurement of Pollution in The Troposphere (MOPITT) multispectral CO retrievals in the Community Earth System Model (CESM). The focus is to assess the impact on the chemical system, i.e. the CH 4 -O 3 -CO-NO X -OH interactions, when CO distribution is constrained by a coupled full chemistry‐climate model like CESM. We show that increases/decreases in CO lead to net reduction of OH and subsequent longer lifetime of CH 4 . Since the assimilation of MOPITT observations constrains the global CO burden, which had significantly decreased over this period by ~20%, we find that this leads to (a) an increase in CO chemical production, (b) a higher CH 4 oxidation by OH, and (c) ~8% shorter CH 4 lifetime.

Second, we compare a suite of state-of-the-art global CO 2 inverse models to observations to assess the dependence on differences in northern extratropical vertical transport and to identify other drivers of the modelled spread. The posterior CO 2 concentration profiles have been evaluated against the High-Performance Instrumented Airborne Platform for Environmental Research (HIAPER) Pole-to-Pole Observations (HIPPO) aircraft campaign over the mid pacific in 2009-2011. The modelled CO2 fields agree reasonably well with the HIPPO observations, in particular for the annual mean vertical gradients in the northern hemisphere. The latitudinal distributions of land fluxes have converged significantly since the Atmospheric Carbon Cycle Inversion Intercomparison (TransCom3) and the Regional Carbon Cycle Assessment and Processes (RECCAP) and they are now in close agreement. The results from these models for other time periods (2004-2014, 2001-2004, 1992-1996), and studies results confirm that the tropics have been almost neutral for several decades. However, models do still disagree on the ocean-land partitioning, and this is driven by differences in fossil fuel emissions associated with differences in retrieved atmospheric growth rates.