The long-term record of atmospheric CO2 increase shows that only about 50% of CO2 emitted by fossil fuel burning remains in the atmosphere, with the rest being absorbed by the oceans and terrestrial biosphere. Yet the persistence of this carbon sink is uncertain, with coupled carbon-cycle climate models in Earth system models (ESMs) exhibiting divergent behavior. Most of the model uncertainty appears to be related to differing parameterizations of carbon uptake and its sensitivity to environmental drivers like temperature and moisture. In this proposal, we aim to use a 30 year-long, globally distributed records of CO2 and its 13C:12C ratio (expressed as d13C) to begin to constrain the relationship between drought and carbon uptake. The relationships we observe via atmospheric measurements will be compared to the same relationships in carbon cycle models commonly used in ESMs, thus allowing us to test the validity of their carbon-climate linkages. In particular, we will use a newly developed dual-tracer variant of the CarbonTracker data assimilation system, CTDAS-C13, to use CO2 and d13C data to optimize values of terrestrial net flux (NEE) and isotopic discrimination during photosynthesis (Dph). Dph is related to stomatal conductance, and thus a sensitive indicator of a plant�??s or ecosystem�??s response to drought. Recent work by Peters et al.  showed a strong relationship between NEE and Dph and additionally demonstrated that no biosphere model (out of six tested) exhibited even half the sensitivity of the data-constrained relationship. Here we propose to extend that analysis to more models, a much greater time span and examine the relationship between NEE, Dph and additional variables. As part of the project we will re-examine and improve numerous aspect of the CO2 and d13C budgets required for accurate modeling, including fluxes and isotopic parameters associated with fossil emissions, and oceanic, and terrestrial C exchange. We will also create a new harmonized long-term d13C data set that, along with all model input fluxes, and the CTDAS-C13 model itself, will be made freely available.