Predictability of the Carbon-Climate System on Seasonal to Decadal Time Scales
Project Team
Principal Investigator
Coupled carbon-climate models show that terrestrial carbon processes greatly intensify future global warming compared to uncoupled climate models (without terrestrial carbon processes) even for approximately the same CO2 trajectories. Thus, the predictability of the coupled carbon-climate system may be different from the predictability of the physical climate system alone. Here, we propose to improve our understanding of the predictability of the coupled carbon-climate system, especially terrestrial carbon processes, on seasonal to decadal time scales. We will determine the predictability of terrestrial carbon processes, and will identify the optimal initialization strategies for forecast horizons ranging from seasonal to decadal time scales in the coupled carbon-climate system. We will further identify the variables in the initial conditions that have most impact on the predictability of the terrestrial-carbon related variables on time scales ranging from seasons to decades. The pilot study on the predictability of ecosystem production and terrestrial carbon flux will lay a foundation for the actual prediction, and future comprehensive climate data assimilation and deployment of an observation network designed to improve the prediction of ecosystem production, net terrestrial carbon flux and CO2 concentration. Lack of ensemble reanalysis products to initialize ensemble climate forecasts motivates us to propose to generate the first ensemble atmospheric reanalysis products for meteorological variables and CO2 vertical profiles with the Local Ensemble Transform Kalman filter (LETKF). Our proposed work will provide first-time carbon-climate ensemble reanalysis products, including both traditional meteorological variables and CO2 vertical profiles, and will be a first step toward comprehensive climate data assimilation. The ensemble reanalysis products could be directly used as ensemble initial conditions for the atmosphere. In the proposed study, we will use the NCAR coupled carbon-climate model with either prescribed ocean boundary forcing (generating atmospheric reanalysis products) or interactive ocean (predictability study).
Our pilot study will have a significant impact on the future initialization of ensemble climate forecasts. We will share the carbon-climate ensemble reanalysis products generated in this study with the research community at large, which will impact the initialization of atmospheric states in future ensemble climate forecasts. Our study is pioneering a new strategy to study the predictability of the prediction metric of interest. It will guide the initialization strategy for the prediction of terrestrial carbon processes ranging from seasonal to decadal temporal scales. Finally, our project will train a new generation of researchers (a graduate student and a postdoc) who will gain expertise in both climate data assimilation and climate forecasts.