Each shown spectrum is created from possibly multiple data points selected in "Time range selection" plot. Spectra for each sensor are averaged, and the results are then normalized.

Similarity plots show if acoustic spectra obtained at different points of time are similar. The horizontal axis shows when one spectrum was obtained, the vertical axis shows when the other spectrum was obtained. The larger the distance, the greater is the difference. The distance here is similar to the distance between points on a geographic map. Four different metrics represent four ways of distance calculation. The simplest of them is the Euclidean distance, which is calculated using the Pythagorean theorem, but in a multidimensional space. As you can see, the main diagonal is always blue, indicating zero distance. This happens because main diagonal shows comparison of spectra with themselves. For any true metric, the distance between point A and itself must be zero: Distance(A, A) = 0.

Also, each plot is symmetric with respect to the main diagonal — you can place a mirror along the blue diagonal line and recreate the second half of each plot. This happens because the Distance(A, B) corresponds to Distance(B, A) after reflection. For any true metric, the distance between point A and point B must be equal to the distance between point B and point A: Distance(A, B) = Distance(B, A).

So, what can we see in these plots? One should look for patterns. A periodic pattern will indicate periodicity of a signal over time. For example, a pattern that nearly reproduces itself if shifted horizontally and/or vertically by 24 hours reveals the daily periodicity of a signal.

Keep in mind that scales can be different for different subplots.

For sensor 109:
First datapoint: 2025-06-26 07:52:59+03:00
Last datapoint: 2025-06-28 07:07:48+03:00
Number of datapoints: 51

For sensor 116:
First datapoint: 2025-06-26 08:05:44+03:00
Last datapoint: 2025-06-28 07:24:32+03:00
Number of datapoints: 51

For sensor 117:
First datapoint: 2025-06-26 07:56:26+03:00
Last datapoint: 2025-06-28 06:39:13+03:00
Number of datapoints: 49