Hybrid residual moveout error estimation

A method for estimating RMO error, comprising:

• obtaining production data comprising a domain of picking locations;
• selecting a training data set from the production data;
• and
• training a machine learning system by:
• running a plurality of residual moveout (RMO) error picking algorithms with the training data set;
• for each of the plurality of RMO error picking algorithms, flattening gathers based on the respective RMO error picking algorithms to generate a resultant training error pick;
• extracting features from each of the resultant training error picks;
• comparing the resultant training error picks from the plurality of RMO error picking algorithms at each picking location in the domain of picking locations;
• based on the comparing, assigning a label representative of the resultant training error picks at each picking location;
• and
• calculating hyperparameters at each picking location based on the extracted features and assigned labels.

A machine learning system for estimating RMO error, comprising:

• a processor configured to:
• obtain production data comprising a domain of picking locations;
• select a training data set from the production data;
• and
• train the machine learning system by:
• running a plurality of residual moveout (RMO) error picking algorithms with the training data set;
• for each of the plurality of RMO error picking algorithms, flattening gathers based on the respective RMO error picking algorithms to generate a resultant training error pick;
• extracting features from each of the resultant training error picks;
• comparing the resultant training error picks from the plurality of RMO error picking algorithms at each picking location in the domain of picking locations;
• based on the comparing, assigning a label representative of the resultant training error picks at each picking location;
• and
• calculating hyperparameters at each picking location based on the extracted features and assigned labels.