The space-time variations of the carbon budget at the Earth’s surface are highly variable and
quantifying them represents a major scientific challenge. A prominent strategy consists in
inferring the carbon surface fluxes from the gradients of the carbon concentrations in the
atmosphere. This method has been established in the 90s based on the (sparse)
atmospheric measurements from conventional surface networks. With the advent
of atmospheric composition remote sensing in the 00s, this method can now be
applied to tentatively monitor the carbon surface fluxes over long periods from
space.
For CO2 flux inversion, the requirements on the relative accuracy of the atmospheric
products, like those from AIRS and GOSAT, are particularly stringent. Similarly,
uncertainties in atmospheric transport modeling also hampers the inversion of the carbon
fluxes. Satellite products with little or no sensitivity to the boundary layer, like those of AIRS,
combine the difficulty of a small signal to capture and that of a large contribution of the
transport model needed to interpret them in terms of surface fluxes. The generation of
products from CO2-dedicated instruments, like GOSAT, with some sensitivity to the
boundary layer, is still at an early stage and is complicated by the ambiguity of the measured
short-wave infrared radiation with respect to CO2, aerosols and thin cirrus clouds at
once.
Within the European GEMS project and its follow-on MACC, LSCE has been developing
a variational inversion scheme to extract the information from the satellite sounders about the
surface fluxes. This system is one of the few that processes satellite soundings individually
(i.e. large observation vector), while preserving the high spatial resolution of the expected
inversion flux increments (i.e. large state vector). It has been applied to products
from AIRS and from GOSAT and to the conventional surface measurements for the
estimation of time-varying CO2 surface flux maps at the global resolution of 2.5°Ã3.75°
(latitude, longitude). In this paper, we describe the inversion system and show the
comparison and the evaluation of the three sets of inverted fluxes. Complementary
comparisons are provided from the reference surface network of Total Carbon Column
Observation Network (TCCON) and from a state-of-the-art and observation-free model
simulation of CO2. It is shown that the model simulation is already in agreement
within 1-2 ppm with individual TCCON retrievals at TCCON sites. This result
sets the bar high for satellite CO2 products in general. Our evaluation indicates
that the satellite products tested still have not outperformed the surface network
so far and do not bring new insights about the carbon fluxes yet. We conclude by
synthesizing the strengths and weaknesses of the current satellite observing systems for
surface flux monitoring and make some recommendations for future developments. |