Cross-correlation is a statistical measure of similarity between two inputs as a function of time shifts.
This process computes the cross-correlation between a volume and another volume, trace, or wavelet. Use this process to determine cross-correlation, repeating signals (multiple) or to quantify phase and time differences between input traces.
- In the Control Panel, open the Process tab.
- At the tab header, click the Add icon and select New Process.
- Scroll down and double-click on Cross Correlation.
- Type a name for the process, and click OK.
- Click Close.
- In the Details Panel, select the input Volume.
- Headers: Output the trace shift that would result in the best correlation (in ms, metres, or feet) and/or the best correlation value (0-1000) in selected headers. The input sample data is left unmodified.
Correlogram: Output a correlogram of cross-correlations, each sample corresponding to a particular shift.
- Normalise: When selected, normalises the correlogram.
- For Surface Consistent Processing: Output an attribute map volume with two layers: the shift that would result in the best correlation (in ms, metres, or feet) and the best correlation value (0 - 1). This output can be used as input to SC processing.
- Sliding window: Output a trace containing the best shift (in ms, metres, or feet) or correlation value from the window centred at each output sample.
- Matching trace for each input trace: Cross-correlate each trace in the primary volume with a corresponding trace from the reference volume.
- Single gather: Cross-correlate each trace in the primary volume with a corresponding trace from the reference gather. The reference volume must contain a single gather or record.
- Single trace per gather: Cross-correlate each gather in the primary volume with a corresponding trace from the reference volume. The reference volume must contain a single trace per gather location.
- Single trace: Cross-correlate each trace in the primary volume with the trace in the reference volume.
- Wavelet: Cross-correlate each trace in the primary volume with the specified wavelet.
- Select the Reference Volume/Wavelet.
- Wavelets must be a single wavelet and not a multi-wavelet field.
# of lags (half window): The number of relative shifts to cross-correlate.
- If producing a correlogram, the correlogram will be 2N + 1 samples long, where N is # of lags. I.e. N negative shifts, the zero shift, and N positive shifts. The zero shift is at time=0. Negative shifts are output at negative times and positive shifts are output at positive times.
- input volume: Apply a window to the input volume before correlating.
- reference volume: Apply a window to the reference volume or wavelet before correlating.
- For Windowed trace window:
- Start of window: The start time/distance (or horizon) of the window.
- Length of window: The length of the input volume window.
An example of cross correlation configuration to output the correlogram of two stack volumes is shown below.
The resulting process volume is a correlogram of cross-correlations, each sample corresponding to a particular shift.
The output volume can be used to indicate how good is the correlation between the input and reference volume. If input and reference volume are identical, it will show the highest correlation value centered at 0 ms.
- Picture example: 2D line section view of correlogram output showing a strong event centering at 0ms; indicating good correlation between input and reference volume at the current 2D line
If there is a time and/or phase shift between the input and reference volume, the highest correlation value will deviate from 0 ms.
- Picture example: Map view of correlogram output displayed at 0 ms time slice, indicating areas with high correlation (purple/pink) and areas with lower correlation (grey).
An example of cross correlation configuration to output the time shift that would result in the best correlation between two stack volumes is shown below.
The time shift is stored in the user chosen header; i.e., the samples in the output volume will be identical to the input volume.
To view the results stored in the output header; use attribute map. Select the attribute source to use the output header from cross correlation process (shift and correlation).
If attribute map process is not available, a dummy volume can be created to display the time shift and correlation header value on map view
- Use volume maths to transfer the header value from the cross correlation process to volume sample using formula as shown in the below picture.
Note: “gx” in the formula is the header where the time shift is stored in the input volume.
- In the map view, select the volume and enable the time slice to display the time shift map.
Note: All time slice will display the same information
- Change the header in the volume maths formula to reflect the cross correlation header to display the cross correlation map instead of the time shift map.
To generate a phase shift map based on peak amplitude time shift, see Generating spatially variable phase shift map between two stack volumes.