Fig. 4: Reduced gap of the annual total anthropogenic methane of ensemble inverse estimates and inventory emissions in Alberta and Saskatchewan; however, a significant discrepancy persisted prior to the 2024 NIR major methodological updates.

The inverse estimates for the four atmospheric transport models are shown here. Average results incorporating the three prior emission scenario inputs, two inversion methods and two cold periods of Nov–Dec and Jan–Mar are shown using the 7-sub-region inversion setup. 0.2 Mt of wetland emissions have been subtracted. The error bars show the ±1 standard deviation (STD) of ensemble estimates by transport model and by year. The red horizontal line shows the 5-year average of 2010–2014 for the ensemble inverse methane estimates. The light pink shaded band shows the ±1 STD of the ensemble results over the 2010–2022 period. The 5-year average of inventory emissions for the 2023, 2022 and 2021 NIRs are shown as blue, gray and black horizontal bars respectively. The red and blue dashed lines show the 2010–2022 averages for the inverse and 2024 NIR methane emission estimates. ERA5-FP: The ERA5 reanalysis meteorological data from the European Centre for Medium-range Weather Forecasts driving the FLEXPART dispersion model. GDASD-HS: the Global Data Assimilation System data from the National Weather Service’s National Centers for Environmental Prediction driving the US. NOAA’s HYSPLIT dispersion model. GEM-FP: the global environmental multi-scale model from ECCC driving the FLEXPART dispersion model. WRFaN-HS: the Advanced Research Weather Research and Forecasting time-averaged meteorological data driving the US. NOAA’s HYSPLIT dispersion model. For details, see Supplementary Method 3.1.