Surrogate Gas Prediction Model as a Proxy for {Δ}14C-Based Measurements of Fossil Fuel-CO2

Published

June 27, 2016

Author(s)

Kevin J. Coakley, Benjamin Miller, Stephen A. Montzka, Colm Sweeney, Ben Miller, John B. Miller

Abstract

In contrast to CO2 produced in the atmosphere by all other sources, CO2 produced by the combustion of fossil fuels is devoid of 14C. Thus, the measured 14C: 12C isotopic ratio of atmospheric CO2 (and its associated derived {Δ}14C value) is an ideal tracer for determination of the fossil fuel derived CO2 enhancement contributing to any atmospheric CO2 measurement (Cff). Given enough such measurements, independent, top-down estimation of US fossil fuel- CO2 emissions should be possible. However, the number of {Δ}14C measurements is presently constrained by cost, available sample volumes, and accessibility to high precision accelerator mass spectrometer measurement facilities. {Δ}14C is therefore measured in just a small fraction of samples obtained by flask air sampling networks around the world. Here, we develop a Projection Pursuit Regression model to predict Cff as a function of multiple surrogate gases acquired within the NOAA/ESRL Global Monitoring Division Greenhouse Gas Reference Network. The surrogates consist of measured enhancements of various anthropogenic trace gases, including CO, SF6, and halo- and hydro- carbons acquired in vertical airborne sampling profiles near Cape May, NJ and Portsmouth, NH from 2005 through 2010. We select the complexity and form of the Projection Pursuit Regression model with a statistical learning method called cross validation. Model performance is quantified based on predicted values corresponding to test data excluded from the model building process. The root-mean-square difference between predicted and observed Cff (expressed in amount of substance fractions) for various definitions of the test data varied from 1.06 to 1.37 υmol/mol (ppm). In contrast, the component of uncertainty of any {Δ}14C-based measurement due to random effects is 1 ppm. If a proxy prediction model based solely on measurements of surrogate gases were to perform comparably well throughout the domain of the US portion of the ...

Citation

Journal of Geophysical Research

Volume

121

Issue

12

Pub Type

Journals

Download Paper

https://doi.org/10.1002/2015JD024715

Keywords

aircraft air samples, anthropogenic trace gases, C14:C12 ratio, delta 14C, fossil fuel CO2 emissions, model selection, projection pursuit regression, proxy prediction model, statistical learning

Citation

Coakley, K. , Miller, B. , Montzka, S. , Sweeney, C. , Miller, B. and Miller, J. (2016), Surrogate Gas Prediction Model as a Proxy for {Δ}14C-Based Measurements of Fossil Fuel-CO2, Journal of Geophysical Research, [online], https://doi.org/10.1002/2015JD024715 (Accessed April 19, 2024)