Аннотация:A new method for isolating the quasi-periodiccomponent of a time series based on the description of its form is proposed. This form is set by alternating convex up and convex down sections, the inflection points are the form parameters. The quasi-periodic component is isolated by solving the problem of the best approximation of the presented series by quasi-periodic signals.This approach makes it possible to distinguish a componentwith a variable period in the time series. After morphological filtration of the component of the series modeling the daily variability, the remainder of the series becomes stationary, which allows using methods of mathematical statistics and Fourier analysis for its further study. Verification of the obtained results was carried out by comparison with the results of Fourier analysis. The effectiveness of the approach is illustrated by results of decomposition of a time series of CO2 concentration in the atmosphere.