AGWS curve shows the averaged global wavelet spectrum computed for all time series of the space-time diagram. It shows the most dominant periods found in the entire space-time diagram. This curve informs us about (quasi-)periodical processes in our data set. For an example - see H. Meszarosova and P. Gomory, A&A 643, A140 (2020), Apendix A.

Detecting and separating of the the hidden structures B and C: a) The global wavelet spectra GWSs for all of the individual SUM time series were computed and integrated to get an averaged global wavelet spectrum AGWS1. This curve displays the most dominant periods (curve peaks), which were found in the space-time diagram SUM.

We selected a limit L1 of 76 s (according to the AGWS1 peak) for the data separation. The new separated space-time diagrams were computed for the period ranges A2 = PR > L1 and PR < L1. Practically, we selected only time series of the individual period range PR from the diagram SUM to put them into the new separated diagram. Thus, for example, new diagram A2 only consists of the time series with PR > L1. Now the strong continuum is separated in the new A2 space-time diagram.

To prove this, we selected one time series (as an example) to compute its wavelet power spectrum. We gained a good separation since the event A2 = A; except for the left and right narrow margins of the separated A2 time series, because of an eect of finite time series length.

Similarly, the global wavelet spectra (GWSs) for all of individual time series of the PR < L1 diagram were computed and integrated to obtain new averaged global wavelet spectrum AGWS2. Then we selected a limit L2 = 12 s for data separation with the period ranges. The global wavelet spectra GWS enable us to determine a characteristic period P of 60 and 9 s for the events B1 and C1, respectively.

The resulting space-time diagram SUM is a summary of these events A1 + B1 + C1, where the events B1 and C1 are invisible and hidden below event A1. This is a common occurrence in observed data.

The GWS curve (Torrence, C., & Compo, G. P. 1998, Bull. Am. Meteorol. Soc., 79, 61) displays a global wavelet spectrum computed for the selected time series to determine a characteristic period P above a wavelet significance level of >95%.