Investigating high-resolution methane emission uncertainty reduction in Gippsland using in-situ data: A Bayesian inverse modeling and variational assimilation approach

Quantifying methane emissions in Gippsland, Victoria, Australia is challenging due to the presence of multiple emission sources, resulting in overlapping emissions and considerable uncertainty in estimation. To address this challenge, our study investigates the potential to reduce uncertainties in methane emissions in Gippsland through the combination of in-situ data, models, and prior information using a Bayesian inverse modeling and variational approach. We employ a four-dimensional variational in-situ data assimilation technique built around the Community Multiscale Air Quality (CMAQ) model at 2 km resolution for four months in 2019.

Initially, we used the Emission Database for Global Atmospheric Research (EDGAR) as a baseline but identified a number of shortcomings in capturing local emissions. To address this issue, we introduced prior estimates from the "openmethane" prior at https://openmethane.org/. We evaluated the underlying Weather Research and Forecasting Model (WRF) meteorological predictions against nearby weather station data, revealing good performance at most times. We validated the performance of our concentration model by comparing it with observational data at the three sites used in the study.

We will discuss the results and present the reductions in emission uncertainties. Next steps in the study will integrate these findings to further rectify biases and improve the accuracy of methane emission estimates in the Gippsland region, especially during the intense fire period of 2019-2020.