Autocorrelation
Autocorrelation identifies periodicity in a signal, e.g. air gun bubble, reverberations and multiples. It is useful for detecting repeating periods within signals in the presence of noise.
At each lag value, the signal is correlated with a time-shifted copy of itself. The autocorrelation result is often normalised such that the value at zero lag is 1.
Create Autocorrelation process
![Create autocorrelation process](https://media.screensteps.com/image_assets/assets/003/226/621/original/e3af5583-3e85-49e7-80f1-8ff7b8ed887d.png)
- In the Control Panel, open the Process tab.
- Click the blue "+" icon and select New Process.
- Search and double-click Autocorrelation.
- Type a name for the process and click OK.
Define Autocorrelation settings
![](https://media.screensteps.com/image_assets/assets/001/373/482/original/c3a36902-e358-48f7-b48c-7846281dd87c.png)
- Volume: Select the data to autocorrelate.
-
# of lags: The number of correlations (shifts by sample interval) to compute in the autocorrelogram.
- This includes the zero shift. e.g. "# of lags" = 3 gives an autocorrelogram with lags of 0, 1 and 2 samples).
- The number of lags determines the output trace length in samples.
- The zero lag will be at time 0.
- Normalise: Normalise the result so the zero lag autocorrelation is 1.
-
Trace windowing: Calculate the result for the full trace or a window
-
Windowed
- Start of window: The start time/distance or horizon of the window.
- Window Length: The length of the input volume window.
- Entire Trace
-
Windowed
- As a result of this process, a new volume is available in the Volume tab.
Example of autocorrelation
Nice example of a strong zero phase event at 0 ms with little noise in the data below. No coherent events below suggesting residual noise was removed.
![Example of autocorrelation](https://media.screensteps.com/image_assets/assets/001/373/480/original/421a9f01-3821-43cb-bd1f-4427e205366c.png)