Observing Chinese Coal Mine Methane Emissions Smoothly Across Scales: Powering Future Mitigation

This work describes observations made over the past three years at high gas coal mines in Shanxi and Xinjiang observed from underground; surface concentrations and fluxes; horizontal and upward looking FTIR; high spatial resolution remote sensing using moderate spectral resolution (GF5 and PRISMA); and lower spatial resolution remote sensing using high spectral resolution (TROPOMI). Systematic analysis is made using flexible, mass-conserving, and computationally fast inversion tools. High resolution emissions are computed and analyzed considering spatial variation, wind, three-dimensional spread, and observational uncertainty. These emissions are observed to contain fat tail distributions and uncertainties. When applied in attribution forward-backward mode, a probabilistic attribution is computable, with results consistent between different observations tyles when and where they were made. Attribution is sometimes possible both for the known sources and other second source (i.e., adjacent mines, additional fissures or ventilation shafts, or long-range transport from outside the domain). Advantages, weaknesses, and ranges of uncertainties of each observation type is explored.

These results are then used to train mesoscale mass conservation systems that govern the transport, diffusion, and interactions, allowing for emissions estimation, uncertainty, and attribution. These trained models are applied to TROPOMI and other observations at the kilometer to 10-kilometer scale, and demonstrate day-by-day and grid-by-grid emissions which are quantifiable and reasonable when compared with independent surface observations. Furthermore, the results are shown to be smooth and consistent across some other coal mining areas, when and where observational uncertainties were initially strictly considered in the model fitting and after unbiased analysis is applied.

Issues of when the errors are large or emissions estimations are not reliable are discussed including: different coal fields, underground coal fires, high absorbing aerosol conditions, and variable topography. When emissions can be computed under these conditions, reasons are given, and future work options are discussed. New measurements and campaigns and modeling enhancements will be discussed. In specific, limitations on the current generation of surface and remotely sensed measurements will be made in terms of ever tightening emissions rules. Steps to rectify the identified gaps are proposed.

General results reflect current understanding: high gas mines are significant sources of methane emissions which require active mitigation or yield emissions larger than current bottom-up estimates. Treating emissions as normally distributed leads to results not being sufficiently robust to extrapolate to annual or longer datasets. Some scientific points raised include: active consideration of observational error frequently leads to emissions inversions not being reliable; applying unbiased filters to observational uncertainty removes both high and low emissions inversions, allowing more confidence in unfiltered low emissions sources; attribution of multiple sources on a single TROPOMI grid may be easier when applied over a common coal field, allowing field wide emissions predictions under lower-gas or higher mitigation conditions.