Data Assimilation for Atmospheric CO2: Towards Improved Estimates of CO2 Concentrations and Fluxes.

Carbon Cycle Data Assimilation;CO2 Flux Estimation;Ensemble Kalman Filter;4 Dimensional Variational (4DVAR);GOSAT;Error Covariance Estimation;Civil and Environmental Engineering;Atmospheric;Oceanic and Space Sciences;Geology and Earth Sciences;Natural Resources and Environment;Statistics and Numeric Data;Engineering;Science;Environmental Engineering

Chatterjee, AbhishekPosselt, Derek J. ;

University of Michigan

关键词: Carbon Cycle Data Assimilation;    CO2 Flux Estimation;    Ensemble Kalman Filter;    4 Dimensional Variational (4DVAR);    GOSAT;    Error Covariance Estimation;    Civil and Environmental Engineering;    Atmospheric;    Oceanic and Space Sciences;    Geology and Earth Sciences;    Natural Resources and Environment;    Statistics and Numeric Data;    Engineering;    Science;    Environmental Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/96172/abhishch_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

The lack of a process-level understanding of the carbon cycle is a major contributor to our uncertainty in understanding future changes in the carbon cycle and its interplay with the climate system. Recent initiatives to reduce this uncertainty, including increases in data density and the estimation of emissions and uptake (a.k.a. fluxes) at fine spatiotemporal scales, presents computational challenges that call for numerically-efficient schemes. Often based on data assimilation (DA) approaches, these schemes are common within the numerical weather prediction community. The goal of this research is to identify fundamental gaps in our knowledge regarding the precision and accuracy of DA for CO2 applications, and develop suitable methods to fill these gaps. First, a new tool for characterizing background error statistics based on predictions from carbon flux and atmospheric transport models is shown to yield improved estimates of CO2 concentration fields within an operational DA system at the European Centre for Medium-Range Weather Forecasts (ECMWF). Second, the impact of numerical approximations within existing DA approaches is explored using a simplified flux estimation problem. It is found that a complex interplay between the underlying numerical approximations and the observational characteristics regulates the performance of the DA methods. Third, a novel and versatile DA method called the geostatistical ensemble square root filter (GEnSRF) is developed to leverage the information content of atmospheric CO2 observations. The ability of GEnSRF to match the performance of a more traditional inverse modeling approach is confirmed using a series of synthetic data experiments over North America. Fourth, GEnSRF is used to assimilate high-density satellite observations from the recently launched GOSAT satellite, and deliver global data-driven estimates of fine-scale CO2 fluxes. Diagnostics tools are used to evaluate the benefit of satellite observations in constraining global surface fluxes, relative to a traditional surface monitoring network. Overall, this research has developed, applied, and evaluated a novel set of tools with unique capabilities that increase the credibility of DA methods for atmospheric CO2 applications. Such advancements are necessary if we are to accurately understand the critical controls over the atmospheric CO2 growth, and improve our understanding of carbon-climate feedbacks.