A33F-3253:
Estimating National-scale Emissions using Dense Monitoring Networks

Wednesday, 17 December 2014
Anita Ganesan1, Alistair Manning2, Aoife Grant1, Dickon Young1, David Oram3, William T Sturges3, John B Moncrieff4 and Simon O'Doherty1, (1)University of Bristol, Bristol, United Kingdom, (2)UK Meteorological Office, Exeter, United Kingdom, (3)University of East Anglia, Norwich, United Kingdom, (4)University of Edinburgh, Edinburgh, United Kingdom
Abstract:
The UK’s DECC (Deriving Emissions linked to Climate Change) network consists of four greenhouse gas measurement stations that are situated to constrain emissions from the UK and Northwest Europe. These four stations are located in Mace Head (West Coast of Ireland), and on telecommunication towers at Ridge Hill (Western England), Tacolneston (Eastern England) and Angus (Eastern Scotland). With the exception of Angus, which currently only measures carbon dioxide (CO2) and methane (CH4), the remaining sites are additionally equipped to monitor nitrous oxide (N2O). We present an analysis of the network’s CH4 and N2O observations from 2011-2013 and compare derived top-down regional emissions with bottom-up inventories, including a recently produced high-resolution inventory (UK National Atmospheric Emissions Inventory). As countries are moving toward national-level emissions estimation, we also address some of the considerations that need to be made when designing these national networks.

One of the novel aspects of this work is that we use a hierarchical Bayesian inversion framework. This methodology, which has newly been applied to greenhouse gas emissions estimation, is designed to estimate temporally and spatially varying model-measurement uncertainties and correlation scales, in addition to fluxes. Through this analysis, we demonstrate the importance of characterizing these covariance parameters in order to properly use data from high-density monitoring networks. This UK case study highlights the ways in which this new inverse framework can be used to address some of the limitations of traditional Bayesian inverse methods.