Carbon cycle climate feedback uncertainties play a large role in defining the overall envelope of uncertainty for predictions of future greenhouse gas concentrations, terrestrial ecosystem structure and function, and associated climate changes. The processes thought to dominate the sign and magnitude of carbon-cycle feedback vary across latitudinal zones, and integrated prediction of the global-scale feedbacks depends on detailed understanding of regional and zonal mechanisms connecting the terrestrial cycles of carbon and nutrients with the climate system.

Our objectives in this project are to:

1. Quantify critical uncertainties in global-scale climate predictions associated with carbon-climate feedbacks
2. Improve our understanding and model representation of processes controlling these feedbacks through zonally specific model-data evaluation exercises
3. Extend our data-based evaluation to a quantification of carbon-climate feedback responses and uncertainties in a large population of global-scale models.

We have prioritized our efforts to focus on three of the most significant terrestrial ecosystem types, in terms of their contributions to the global carbon cycle, categorized by latitudinal zones:

1. High-latitude forest, shrub, and tundra systems, with a focus on permafrost systems (effort led by W.J. Riley)
2. Tropical upland and lowland forests, with a focus on carbon-nutrient interactions (effort led by P.E. Thornton)
3. Forests of the temperate zone, with a focus on the consequences of land use modification and age-class dynamics over the past several centuries (effort led by R.A. Fisher).

In each of these zones and ecosystem types, global model representations of carbon cycle-climate interactions, as well as interactions with water and nutrient cycles, are limited at present both by a lack of fundamental understanding of relevant processes, and by inadequate synthesis and incorporation of available observational and experimental work. The approach outlined below is designed to overcome these limitations to deliver model predictions with reduced carbon-climate feedback uncertainty, as well as provide an improved framework for evaluating the among-model uncertainty in carbon-climate feedback properties of climate change predictions. Our goal is to reduce the range of climate prediction uncertainty through improved insights into relevant feedback processes and identification of model predictions inconsistent with available observations and experimental results.