Interannual variability in ocean and land CO2 fluxes derived using Time Dependent Inverse model

Using a Bayesian Time Dependent Inverse (TDI) model sources and sinks of CO₂ from 64 regions of the globe have been derived from the atmospheric data at about 100 stations. This inverse model version is primarily based on the TransCom-3 protocol for the distribution of monthly pulse functions of regional sources, and four background fluxes. The NIES/FRSGC global transport model has been run on the Earth Simulator to distribute the monthly and yearly tracer pulse functions at 15 sigma vertical layers and 2.5⁰ x 2.5⁰ horizontal resolution. The model transport is driven by 6-hourly interannually varying meteorological fields from the NCEP/NCAR reanalysis dataset. The target period of our study is set to January 1988 to December 2001. Overall the estimated fluxes are in fairly good agreement with previous estimates (direct and inverse) at both regional and global scales. The average flux rates for the global land and ocean regions are estimated to be -0.8+/-1.2 (1σ ), and -1.9+/-0.8 Pg C/yr, respectively for the period of our study (excluding the 1997/1998 El Niño period). Since the oceanic CO₂ fluxes are relatively undisturbed by the human influences, regional anomalies are closely linked with the major modes of climate variabilities. A detailed correlation study indicates that the flux anomalies have mode of variabilities that are distinct from region to region; e.g., the flux anomalies in northern most Pacific Ocean is linked to the PDO, the anomalous component of Eastern Pacific and Southern Ocean fluxes covary with ENSO/SOI, while the Northern Ocean flux anomalies are controlled by the NAO/AO variability. Details of the land and ocean region fluxes will be presented in the meeting.