Abstract. Observed variations of the atmospheric greenhouse gas methane (CH₄) over the past two decades remain the subject of debate. These variations reflect changes in emission, uptake, and atmospheric chemistry and transport. We isolate changes in the seasonal cycle of atmospheric CH₄ using a wavelet transform. We report a previously undocumented persistent decrease in the peak-to-peak amplitude of the seasonal cycle of atmospheric CH₄ at six out of seven high northern latitude sites over the past two to three decades. The observed amplitude changes are statistically significant for sites at Barrow, Alaska and Ocean Station M, Norway, which we find are the most sensitive of our sites to high northern latitude wetland emissions. We find using a series of numerical experiments using the TM5 atmospheric chemistry transport model that increasing wetland emissions and/or decreasing fossil fuel emissions can explain these observed changes, but no significant role for trends in meteorology and tropical wetlands. We also find no evidence in past studies to support a significant role for variations in the hydroxyl radical sink of atmospheric CH₄. Using the TM5 model we find that changes in fossil fuel emissions of CH₄, described by a conservative state-of-the-science bottom-up emission inventory, are not sufficient to reconcile observed changes in atmospheric CH₄ at these sites. The remainder of the observed trend in amplitude, by process of elimination, must be due to an increase in high northern latitude wetland emissions, corresponding to an annual increase of at least 0.7 %/yr (equivalent to 5 Tg CH₄/yr over 30 years).