A53A-0138: Disaggregating Fossil Fuel Emissions from Biospheric Fluxes: Methodological Improvements for Inverse Methods
Authors: Vineet Yadav1, Yoichi P Shiga2, Anna M Michalak1
Author Institutions: 1. Department of Global Ecology, Carnegie Institution For Science, Stanford, CA, USA; 2. Department of Civil and Environmental Engineering, Stanford University, Stanford, CA, USA
The accurate spatio-temporal quantification of fossil fuel emissions is a scientific challenge. Atmospheric inverse models have the capability to overcome this challenge and provide estimates of fossil fuel emissions. Observational and computational limitations limit current analyses to the estimations of a combined “biospheric flux and fossil-fuel emissions”ù carbon dioxide (CO2) signal, at coarse spatial and temporal resolution. Even in these coarse resolution inverse models, the disaggregation of a strong biospheric signal form a weaker fossil-fuel signal has proven difficult. The use of multiple tracers (delta 14C, CO, CH4, etc.) has provided a potential path forward, but challenges remain. In this study, we attempt to disaggregate biospheric fluxes and fossil-fuel emissions on the basis of error covariance models rather through tracer based CO2 inversions. The goal is to more accurately define the underlying structure of the two processes by using a stationary exponential covariance model for the biospheric fluxes, in conjunction with a semi-stationary covariance model derived from nightlights for fossil fuel emissions. A non-negativity constraint on fossil fuel emissions is imposed using a data transformation approach embedded in an iterative quasi-linear inverse modeling algorithm. The study is performed for January and June 2008, using the ground-based CO2 measurement network over North America. The quality of disaggregation is examined by comparing the inferred spatial distribution of biospheric fluxes and fossil-fuel emissions in a synthetic-data inversion. In addition to disaggregation of fluxes, the ability of the covariance models derived from nightlights to explain the fossil-fuel emissions over North America is also examined. The simple covariance model proposed in this study is found to improve estimation and disaggregation of fossil-fuel emissions from biospheric fluxes in the tracer-based inverse models.