First Name:
Dawn
Last Name:
Woodard
Major Department:
Mathematical Sciences
Thesis Director:
Eric Marland
Date of Thesis:
May 2014
Large point sources account for as much as 60% of the carbon dioxide emissions for some countries. Further, in the US one third of all CO2 emissions come from only 311 point sources (power plants, industrial sites, etc.). Because CO2 emissions are seldom measured directly but are generally estimated from related, proxy, and re-purposed data; we also need to understand the uncertainty of these estimates. Simply stated, given a geographic and temporal space on the Earth, what are the CO2 emissions from that space and what is the uncertainty in this estimate? While the US data on large point sources is largely assumed to have no spatial uncertainty, the actual locations of these sources di↵er by 0.84km on average from their reported locations. Analysis also reveals quantifiable trends in the uncertainty based on simple characteristics such as proximity to water sources, reported location within political boundaries, local and population density. This paper presents a metric to quantify spatial uncertainty in point sources based on the results of this analysis, and explains why point source data cannot be described with traditional methods. To incorporate resolution and placement within a grid cell, a Monte Carlo simulation is used to calculate expected values for emissions for each point source. The spatial uncertainty is then derived from the simulation output to give a picture of the potential spatial spread of the emissions. This is output as gridded data at the desired resolution and can then be incorporated into other data products reporting estimated emissions from point sources. Keywords: uncertainty, carbon dioxide, point source, Monte Carlo