1. International Collaboration on Data Assimilation in Terrestrial Carbon Cycle Science CARBON FUSION www.carbonfusion.org

2. The problem  Major uncertainties remain in the global C cycle (±1GtC yr -1 on land/ocean sink strength)  Bottom-up models (e.g. DGVMs) founder due to heterogeneity in process and difficulties in regional and global corroboration  Top-down models (e.g. inversion) use data so aggregated that it is difficult to generate detailed insights

3. The solution  Model-data fusion:  Combine multiple data-sets with modelling  Use appropriately designed models  Take advantage of recent/planned satellite launches  Employ novel data assimilation approaches  Use primary data rather than products to better quantify uncertainty?

4. The partners - UK  CTCD – Centre for Terrestrial Carbon Dynamics  DARC – Data Assimilation Research Centre  CLASSIC – Climate and Land-Surface Systems Interactions Centre  QUEST – Quantifying and Understanding the Earth System  UK Met Office  ECMWF

5. Collaborators  Max-Planck-Institute for Biogeochemistry, Germany (Martin Heimann)  Laboratory for Climate Sciences and the Environment, France (Philippe Ciais, Peter Rayner)  Department of Atmospheric Science, Colorado State University, USA (Scott Denning)  Marine Biological Laboratory, USA (Edward Rastetter)  CSIRO Earth Observation Centre, Australia (Mike Raupach, Damian Barrett and Pep Canadell)

6. Objectives 1.to accelerate the development of DA in terrestrial C cycle science; 2.to foster international links and knowledge sharing; 3.to develop productive international collaborations; 4.to develop a global vision and strategy for research, in which the UK can fully participate and take a leading role.

7. Overarching science question  How can data assimilation contribute to improving the attribution of terrestrial C fluxes, and to reducing errors in estimating the current state of the C cycle?

8. Subsidiary questions  What are the relative advantages of different approaches to data assimilation?  What data and technical constraints are holding back the science?  How can we best incorporate remote sensing (especially long time-series for C-cycle re-analyses and C column content measurements) into DA schemes?  What is the potential for carbon-water connections in multiple-constraint data assimilation?  How can we assess errors and their propagation through DA schemes?

9. Proposed activities 1.Enhancing research capabilities by exchanges between participating labs 2.Conference and follow-on workshops for discussion and demonstration of current capabilities 3.A special issue focussing on the application of DA to the terrestrial C cycle 4.A research strategy document detailing a vision for the global community 5.A web-page to provide an interactive environment for C cycle research via model-data fusion

10. Today’s objectives  What is the current UK capability in model-data fusion for C cycle science?  What can we learn from other scientific communities on model-data fusion?  What are our current weaknesses and from whom can we learn?  What are the major issues to be addressed in the next 3-5 years?  What key outputs should we plan to produce?