Method for identifying subsurface features
A method for improving a backpropagation-enabled process for identifying subsurface features from seismic data involves a model that has been trained with an initial set of training data. A target data set is used to compute a set of initial inferences on the target data set that are combined with the initial training data to define updated training data. The model is trained with the updated training data. Updated inferences on the target data set are then computed. A set of further-updated training data is defined by combining at least a portion of the initial set of training data and at least a portion of the target data and associated updated inferences. The set of further-updated training data is used to train the model. Further-updated inferences on the target data set are then computed and used to identify the occurrence of a user-selected subsurface feature in the target data set.
1. A method for improving a backpropagation-enabled process for identifying subsurface features from seismic data, the method comprising the steps of:
(a) providing a backpropagation-enabled model, the backpropagation-enabled model having been trained with an initial set of training data comprising seismic data and initial labels identifying a subsurface feature in the seismic data;
(b) providing a target data set;
(c) computing a set of initial inferences on the target data set using the backpropagation-enabled model;
(d) defining a set of updated training data by combining at least a portion of the initial set of training data and at least a portion of the target data set and at least a portion of the set of initial inferences associated with the at least a portion of the target data set;
(e) training the backpropagation-enabled model with the set of updated training data;
(f) computing updated inferences on the target data set with the backpropagation-enabled model trained in step (e);
(g) defining a set of further-updated training data by combining at least a portion of the initial set of training data and at least a second portion of the target data set and at least a portion of the set of updated inferences associated with the at least a second portion of the target data set;
(h) training the backpropagation-enabled model with the set of further-updated training data;
(i) computing further-updated inferences on the target data set with the trained backpropagation-enabled model trained in step (h); and
(j) identifying an occurrence of a user-selected subsurface feature in the target data set from the further-updated inferences.
2. The method of claim 1 , further comprising the step of repeating steps (g)-(i) one or more times.
3. The method of claim 1 , wherein the initial set of training data comprises field-acquired seismic data, augmented seismic data, or combinations thereof.
4. The method of claim 1 , wherein the target data set is comprised of seismic data selected from the group consisting of 1D seismic data, 2D seismic data, 3D seismic data, 4D seismic data, 5D seismic data, and combinations thereof.
5. The method of claim 1 , wherein the target data set is unlabeled.
6. The method of claim 1 , wherein the target data set is labeled.
7. The method of claim 1 , wherein step (d) further comprises a step of thresholding.
8. The method of claim 1 , wherein step (g) further comprises a step of thresholding.
9. The method of claim 1 , wherein the backpropagation-enabled process is selected from the group consisting of supervised processes, semi-supervised processes, and combinations thereof.