Improving a priori emission estimates using network observations and a perturbed emissions ensemble: a study in Beijing

Emissions inventories are essential inputs in air quality modelling from global to local scales, but they also represent a major source of uncertainty in the simulation outputs. Uncertainties may be propagated from the underlying data used to compile the emissions inventories or they can arise from the changes in emissions since the baseline time for which an inventory has been established. Various methods have been developed to validate and/or optimise existing emissions inventories, yet they either require specific measurements which are not readily available or high computing power. More importantly, most of these methods can only provide estimates of the total anthropogenic emissions. This study introduces a novel approach that optimises a priori emission estimates by source sector, which are particularly relevant for policy interventions. At the core of this approach is a perturbed emissions ensemble (PEE), constructed by perturbing parameters in an a priori emissions inventory within their respective uncertainty ranges, which are determined on the basis of expert elicitation. This PEE is then input to an air quality model to generate an ensemble of forward simulations. By comparing the simulation outputs with in-situ measurements of pollutant concentrations from a dense network, the initial uncertainty ranges are constrained and a posteriori emission estimates are derived. Using this approach, an a priori emissions inventory compiled for a study area centred around 2 Beijing in 2013 is updated for the year 2016. CO emissions from the residential sector are estimated to be under 3.5x105 Mg in 2016, less than 39% of the a priori emission estimates. Transport sector emissions of CO and NOX in 2016 are estimated to be 0.9-4.4x105 Mg and 1.5-9×104 Mg, respectively, corresponding to reductions by 44-88% and 57-93% from the prior estimates. Also in 2016, 16-26% of the CO and 15-25% of NOX emitted from transport sources occurred during 0-5 am, higher than the corresponding fraction in the a priori emissions inventory. These a posteriori estimates are derived using two sets of network measurements independently as constraints, supporting the robustness of the approach and these findings. Emission estimates for other source sectors may also be updated if appropriate measurements were available. This approach is generalisable for use with other air quality models and widely applicable in other regions of the world to provide timely updates of emissions by source sector. Potential sources of uncertainty in the approach are also discussed.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2021

Author dc:creator

• Yuan, Le

Advisor dc:contributor.advisor

• Archibald, Alex

• air quality modelling
• emissions
• low-cost sensors
• air quality observations

Licence

• https://www.rioxx.net/licenses/all-rights-reserved/

Language dc:language

eng

Author Identifier

0000-0002-4282-0459

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/341679