In the 1980s, three major advances enabled, for the first time, the study of the global biosphere. These advances were: 1) development of a spatially extensive network of high precision trace gas observations, 2) the first global-scale normalized difference vegetation index (NDVI) observations from Advanced Very High Resolution Radiometer (AVHRR) sensors onboard NOAA weather satellites, and 3) the design of new class of global atmospheric models that included flows of atmospheric carbon dioxide. Inez Fung made a fundamental contribution to the third class of advance, pioneering the development of a carbon dioxide tracer in the GISS atmospheric model. This enabled top-down constraints on the global carbon cycle, including identification of a terrestrial carbon sink in northern forests and a quantitative understanding of terrestrial ecosystem contributions to the annual cycle of CO2 at Mauna Loa and other observing stations. Inez's atmospheric modeling was also critical for quantitatively linking the global-scale NDVI measurements with seasonal and interannual dynamics of atmospheric CO2.

The northern terrestrial carbon sink remains a robust feature of the carbon cycle as we understand it today, and this result has been rigorously tested with many different atmosphere models and with massively expanded atmospheric measurements of CO2 from surface stations, aircraft campaigns, and satellites. The persistence of this sink over time is noteworthy, with significant year-to-year variability, yet with an integrated accumulation in the biosphere of over 100 Pg C during the era of the Mauna Loa time series based on the long-term budget from the Global Carbon Project. Here I discuss the compatibility of this integrated carbon sink with inventory and tree ring observations and estimates of land use change. I provide evidence that a more parsimonious solution to the atmospheric carbon cycle with smaller land use losses and smaller rates of uptake in intact natural ecosystems may also be consistent with available observations on decadal to centennial timescale.