Locating underground features with seismic data processing

A system comprising:

• a seismic source for generating seismic signals;
• a plurality of sensors distributed throughout a region;
• a memory comprising instructions;
• one or more computer processors;
• and
• a computer display, wherein the instructions, when executed by the one or more computer processors, cause the system to perform operations comprising:
• capturing, by the plurality of sensors, a plurality of seismic traces associated with the seismic signals generated by the seismic source;
• for multiple sensors, identifying noise within seismic traces generated by the sensors, the noise comprising at least one of active noise and passive noise;
• for multiple sensors, determining values for attributes associated with the identified noise;
• utilizing an unsupervised machine-learning model, based on the values of the attributes, to determine noise characteristics for each sensor;
• grouping the sensors in a plurality of clusters based on the noise characteristics for each sensor;
• for multiple clusters, creating a noise filter based on the noise characteristics of the sensors in the cluster;
• for one or more sensors, applying the noise filter of the cluster of the sensor to the seismic traces of the sensor;
• analyzing the filtered seismic traces to determine a location of an underground property;
• and
• presenting the determined location of the underground property on the computer display.

A non-transitory machine-readable storage medium including instructions that, when executed by a machine, cause the machine to perform operations comprising:

• capturing, by a plurality of sensors distributed throughout a region, a plurality of seismic traces associated with seismic signals generated by a seismic source;
• for multiple sensors, identifying noise within seismic traces generated by the sensors, the noise comprising at least one of active noise and passive noise;
• for multiple sensors, determining values for attributes associated with the identified noise;
• utilizing an unsupervised machine-learning model, based on the values of the attributes, to determine noise characteristics for each sensor;
• grouping the sensors in a plurality of clusters based on the noise characteristics for each sensor;
• for multiple clusters, creating a noise filter based on the noise characteristics of the sensors in the cluster;
• for one or more sensors, applying the noise filter of the cluster of the sensor to the seismic traces of the sensor;
• and
• analyzing the filtered seismic traces to determine a location of an underground property.