Estimating correlations of CO and CO2 surface fluxes

Monday, 15 December 2014
Brad Weir1,2, Steven Pawson1, Lesley E Ott1, Krzysztof Wargan1,3, Jon Nielsen1,3 and Ricardo Todling1, (1)NASA Goddard Space Flight Center, Global Modeling and Assimilation Office, Greenbelt, MD, United States, (2)Universities Space Research Association Greenbelt, Greenbelt, MD, United States, (3)Science Systems and Applications, Inc., Lanham, MD, United States
Increased understanding of the processes controlling biospheric CO2 flux is
necessary for improved climate projections. While satellite CO2 observations
are expected to better constrain surface sources and sinks, separating
biospheric and anthropogenic fluxes is particularly challenging. Correlations
between CO and CO2 can help to differentiate between different types of
fluxes, but a more thorough understanding of these correlations is still
needed. This presentation discusses the estimation of, and the corresponding
uncertainty about, correlations between CO and CO2 surface fluxes in given
regions during given periods of time. The estimates follow from statistical
diagnostics, including those popularized by Desrozies et al. (2005), applied
to the results of model simulations and assimilations of constituent
measurements, notably data from satellites, in situ sites, and the Total
Carbon Column Observing Network (TCCON). Nevertheless, the usefulness of
these estimates is limited by their uncertainty, a property that depends on
the spatial and temporal resolution of the surface fluxes. This work
considers a variety of different divisions of space (e.g., Global, land and
sea, and Transcom regions) and time (e.g., total, seasonal, and monthly) and
examines the conclusions that can be drawn from the correlation estimates.
For example, the correlation coefficient can give some indication of the
combustion efficiency of emissions from different regions, or it may suggest
that the surface fluxes of the two constituents are from two different
processes. This investigation is used to improve the diagnosis of biospheric
flux processes in the GEOS-5 modeling system.