Exploring uncertainty reduction in high-resolution methane emissions in Gippsland through in-situ data: A Bayesian inverse modeling and variational assimilation method

Abstract

The paper investigates to what extent the assimilation of in-situ data over Gippsland, Victoria, Australia reduces uncertainties in methane emission sources on the regional scale. This was examined via a four-dimensional variational data assimilation system using the Community Multiscale Air Quality (CMAQ) transport-dispersion model. To evaluate the posterior error statistics of optimized monthly-mean methane emissions in Gippsland, we carried out a range of observing system simulation experiments. We ran the assimilations based on four selected months in 2019, employing a horizontal grid resolution of 2 km. The observation data are obtained based on three continuous observation instruments in the Gippsland region. As expected, the largest uncertainty reductions occur near observing sites. Also, our findings indicate that using a high-resolution model and in-situ observations provides detailed information on point sources but offers limited insight into area sources. The overall uncertainty for regional fluxes remains largely unchanged. Therefore, in-situ data is crucial for understanding point sources due to its detailed and localized nature. Finally, uncertainty reduction is much larger when the full concentration dataset is used rather than just the daytime data. This suggests the importance of model improvement to allow use of nighttime data, at least under conditions where the transport model can be expected to simulate atmospheric mixing well.