IG3IS Webinar series 2026: Beyond Bias: Objective Assessment of Transport Model Uncertainty for Greenhouse Gas Verification
Speaker:
Dr Dafina Kikaj, Senior Scientist at the National Physical Laboratory, United Kingdom
About the speaker:
Dr Dafina Kikaj is an atmospheric physicist and Senior Scientist in the Greenhouse Gas Metrology Group within the Atmospheric Environmental Science Department at the National Physical Laboratory, UK. She holds a PhD in atmospheric physics with a focus on quantifying the atmospheric mixing state over complex terrain using the radioactive noble gas radon as a tracer. Her current research involves developing new radon-based methods to identify errors in chemical transport models, thereby aiding the evaluation of uncertainties in GHG emission inventories. Dr Kikaj's work is significant in the climate research community, as she presents a standardized radon dataset for two atmospheric GHG monitoring stations, the UK DECC network’s Heathfield site and the ICOS Weybourne site. These datasets enable optimal utilization of radon measurements for climate research.
Overview of the topic:
Atmospheric transport models provide the critical link between measured greenhouse gas concentrations and the emissions that produced them. However, uncertainties associated with atmospheric mixing, boundary-layer dynamics, vertical transport, meteorological inputs, model resolution, and representation of observation sites can strongly influence the resulting emission estimates. This webinar will examine where transport-model uncertainties arise, how they are currently assessed, and how we can move towards more objective and operational approaches. It will explore the use of natural tracers, meteorological observations, and vertical greenhouse gas gradients to identify model limitations, distinguish among different atmospheric mixing conditions, and assign uncertainty more consistently. The talk will also consider situations where these additional observations are unavailable and discuss alternative approaches, including model ensembles, proxy diagnostics, sensitivity analyses, and uncertainty inflation. Finally, it will outline what is needed for the next generation of greenhouse gas verification systems, in which transport-model uncertainty is routinely diagnosed, quantified and propagated into emission estimates. The central question is not only whether a transport model is biased, but whether we can determine when, where, and under which atmospheric conditions its outputs can be trusted.
