*3.3. Characterization of the ANEs*

In previous works, two parameters were considered by the authors to figure out the effects of the ANEs on noise-map generation [52]. The first of the parameters is based on the classical SNR calculation, consisting of the ratio of power of the ANE in relation to the power of the surrounding RTN. The second metric determines the impact of the ANE on the equivalent noise level used to build the noise map. The calculation of the two parameters is described below.

### 3.3.1. SNR Calculation

As aforementioned, the SNR is calculated as the classical signal-to-noise ratio used in signal processing, considering that the ANE corresponds to the signal and the RTN is the noise. The acoustic power of the ANE and the RTN are calculated as follows:

$$P\_{\mathbf{X}} = \sum\_{n=1}^{N} \left( \frac{\mathfrak{x}[n]^2}{N} \right) \tag{1}$$

where *x*[*n*] is the recorded audio with *N* samples that belongs to either the ANE of the RTN.

After the power calculations of the ANE and the surrounding RTN, the SNR is calculated as:

$$SNR = 10\log\_{10}\left(\frac{P\_{ANE}}{P\_{RTN}}\right) \tag{2}$$

where *PANE* belongs to the anomalous event in question and *PRTN* is the power of the surrounding RTN.

All the casuistry of the calculation is detailed in [52]. Finally, it is worth mentioning that the SNR of a particular ANE could be negative if the power of the surrounding RTN is higher than the power of the ANE itself. This may happen in cases where RTN masks other low-energy sounds, e.g., birds, because of the fluctuation of the road pass-bys.
