Assessment of bottom-up greenhouse gas emissions estimates through independent methods is needed to demonstrate whether reported values are accurate or if bottom-up methodologies need to be refined. Previous studies of measurements of atmospheric methane (CH4) in London revealed that inventories substantially underestimated the amount of natural gas CH4 1,2. We report atmospheric CH4 concentrations and δ13CH4 measurements from Imperial College London since early 2018 using a Picarro G2201-i analyser. Measurements from May 2019-Feb. 2020 were compared to the values simulated using the dispersion model NAME coupled with the UK national atmospheric emissions inventory, NAEI, and the global inventory, EDGAR, for emissions outside the UK. Simulations of CH4 concentration and δ13CH4 values were generated using nested NAME back-trajectories with horizontal spatial resolutions of 2 km, 10 km and 30 km. Observed concentrations were underestimated in the simulations by 12 %, and there was no correlation between the measured and simulated δ13CH4 values. CH4 from waste sources and natural gas comprised of 32.1 % and 27.5 % of the CH4 added by regional emissions. To estimate the isotopic source signatures for individual pollution events, an algorithm was created for automatically analysing measurement data by using the Keeling plot approach. Over 70 % of source signatures had values higher than -50 ‰, suggesting large amounts of natural gas CH4. The analyses based on model-data comparison of δ13CH4 and on Keeling plot source signature emission both indicate that emissions due to natural gas leaks in London are being under-reported in the NAEI. These results suggest that estimates of CH4 emissions in urban areas need to be revised in the CH4 emissions inventories. 1 Helfter, C. et al. (2016), Atmospheric Chemistry and Physics, 16(16), pp. 10543-10557 2 Zazzeri, G. et al. (2017), Scientific Reports, 7(1), pp. 1-13