Methane (CH4), a potent greenhouse gas, traps heat in the atmosphere and significantly contributes to global warming. It is unclear whether CH4 emissions from various land-types and other natural sources have increased substantially in the last decade linked, for example, to global warming and uncertainties remain regarding sources and their spatial extent causing discrepancies between emission estimates from inventories/models and estimates inferred by an ensemble of atmospheric inversions. Here we compared remotely sensed CH4 total column data, along with surface albedo from the Sentinel-5 Precursor (S-5p) satellite against six main temperate zone land types (marsh, swamp, forest, grassland, cropland, and barren-land across Canada over a four-year period (2019–2022). The study developed a machine learning based algorithm that can be used to classify between such different land types using S-5p products. From 2019 to 2022, the average producer’s accuracy (PA) across all land types ranged from 50.8 % to 98.4 %, while the average user’s accuracy (UA) ranged from 69.9 % to 95.4 %. Although the methodology presented does not directly differentiate the methane fluxes from different land types, it does provide a foundation that with better ground truth monitoring and higher resolution imagery, could lead to a being able to differentiate methane emissions between land types with increased confidence, as well as determining whether significant changes are occurring over time. This would yield valuable insights for climate scientists and policy makers at both national and international levels.