A 4D-Var CO2 inversion system with NICAM-TM: development and sensitivity analyses

Abstract:

Our understanding of the global carbon cycle and its feedback mechanism to climate changes is limited due to high uncertainties in estimates of regional CO₂ fluxes at the earth surface. Recently, a large amount of CO₂ concentration data are becoming available from high-frequency aircraft measurements (e.g., CONTRAIL) and satellite measurements (e.g., GOSAT and OCO-2), in addition to expansion of surface measurement networks. To exploit those observational data, a new inversion system has been developed with the four-dimensional variational (4D-Var) method. The system is based on Nonhydrostatic ICosahedral Atmospheric Model-based Transport Model (NICAM-TM), which consists of forward and adjoint transport modes. For the a priori fluxes at terrestrial biospheres and oceans, CO₂ flux data from Vegetation Integrative SImulator for Trace Gases (VISIT) and the diagnostic model of Japan Meteorological Agency (JMA) are respectively used. In the iterative calculation, the quasi-Newton method of Preconditioned Optimizing Utility for Large-dimensional analyses (POpULar) is used. In this study, we present the structure of the newly developed system and performances for CO₂ flux estimates in ideal twin experiments. By the twin experiments, sensitivities to prior error covariance, numerical algorithms, and observational networks are investigated.

Acknowledgment: This study is supported by the Environment Research and Technology Development Fund (2-1401) of the Ministry of the Environment, Japan.