Historical geological data for machine-learning

1 . A system comprising:

a processing device; and

a non-transitory computer-readable memory device that includes instructions executable by the processing device for causing the processing device to perform operations comprising:

receiving historical geological data that includes data about an area of interest;

pre-processing the historical geological data by applying a selected pre-processing technique to the historical geological data with respect to time-attributed geological phenomena to generate geological data, wherein the selected pre-processing technique adjusts the historical geological data to indicate a temporal relationship between the historical geological data and the time-attributed geological phenomena by:

determining an age of the historical geological data relative to the time-attributed geological phenomena;

generating the geological data by restricting the historical geological data to a time window based on the age of the historical geological data relative to the time-attributed geological phenomena, the time window being before present day; and

splitting the time window into a plurality of bins, each bin of the plurality of bins comprising a binary value that indicates whether a geological phenomenon occurred in a subset of the time window corresponding with the bin;

training a machine-learning model using the geological data to generate a trained machine-learning model;

applying the trained machine-learning model to the geological data to generate one or more predictions of geological phenomena with respect to the area of interest;

providing a user interface that includes a visualization of the one or more predictions of geological phenomena, the visualization including a representation of the area of interest and an indication of a likelihood that the one or more predictions of geological phenomena exists at one or more locations included in the area of interest; and

outputting a command to control a natural resource operation for forming a wellbore in the area of interest, the command comprising instructions that are executable to adjust a direction of a drill bit in response to receiving the one or more predictions of geological phenomena to alter a trajectory of the wellbore.

2 . The system of claim 1 , wherein the operation of pre-processing the historical geological data includes:

determining, among a plurality of pre-processing techniques, the selected pre-processing technique to use for pre-processing the historical geological data; and

further pre-processing the historical geological data using a subsequent pre-processing technique of the plurality of pre-processing techniques, the subsequent-pre-processing technique being different than the selected pre-processing technique.

3 . The system of claim 2 , wherein the plurality of pre-processing techniques includes an initial relative timing pre-processing technique, an event barcode pre-processing technique, an event histogram pre-processing technique, a closest age pre-processing technique, and a total count pre-processing technique.

4 . The system of claim 1 , wherein the operation of applying the trained machine-learning model to the geological data to generate the one or more predictions of geological phenomena with respect to the area of interest includes applying the trained machine-learning model to determine a likelihood that a particular natural resource deposit has formed at or adjacent to the area of interest, and wherein the particular natural resource deposit includes a hydrocarbon reservoir.

5 . The system of claim 1 , wherein the operation of applying the trained machine-learning model to the geological data to generate the one or more predictions of geological phenomena with respect to the area of interest includes applying the trained machine-learning model to determine a paleo-digital elevation model for the area of interest.

6 . The system of claim 1 , wherein the user interface includes:

a background that includes a spatial definition of the area of interest; and

a plurality of indicators, wherein each indicator of the plurality of indicators is positioned on the background in a location on the user interface that corresponds to a corresponding location of the area of interest, and wherein each indicator of the plurality of indicators visually indicates:

a likelihood of a natural resource deposit existing at the corresponding location of the area of interest; or

a relative orogenic age of the corresponding location of the area of interest.

7 . A method comprising:

receiving, by a computing device, historical geological data that includes geological data about an area of interest;

pre-processing, by the computing device, the historical geological data by applying a selected pre-processing technique to the historical geological data with respect to time-attributed geological phenomena to generate geological data, wherein the selected pre-processing technique adjusts the historical geological data to indicate a temporal relationship between the historical geological data and the time-attributed geological phenomena by:

determining an age of the historical geological data relative to the time-attributed geological phenomena;

generating the geological data by restricting the historical geological data to a time window based on the age of the historical geological data relative to the time-attributed geological phenomena, the time window being before present day; and

splitting the time window into a plurality of bins, each bin of the plurality of bins comprising a binary value that indicates whether a geological phenomenon occurred in a subset of the time window corresponding with the bin;

training, by the computing device, a machine-learning model using the geological data to generate a trained machine-learning model;

applying, by the computing device, the trained machine-learning model to the geological data to generate one or more predictions of geological phenomena with respect to the area of interest;

providing, by the computing device, a user interface that includes a visualization of the one or more predictions of geological phenomena, the visualization including a representation of the area of interest and an indication of a likelihood that the one or more predictions of geological phenomena exists at one or more locations included in the area of interest; and

outputting a command to control a natural resource operation for forming a wellbore in the area of interest, the command comprising instructions that are executable to adjust a direction of a drill bit in response to receiving the one or more predictions of geological phenomena to alter a trajectory of the wellbore.

8 . The method of claim 7 , wherein pre-processing the historical geological data includes:

determining, by the computing device and among a plurality of pre-processing techniques, the selected pre-processing technique to use for pre-processing the historical geological data; and

further pre-processing, by the computing device, the historical geological data using a subsequent pre-processing technique of the plurality of pre-processing techniques, the subsequent pre-processing technique being different than the selected pre-processing technique.

9 . The method of claim 8 , wherein the plurality of pre-processing techniques includes an initial relative timing pre-processing technique, an event barcode pre-processing technique, an event histogram pre-processing technique, a closest age pre-processing technique, and a total count pre-processing technique.

10 . The method of claim 7 , wherein applying the trained machine-learning model to the geological data to generate the one or more predictions of geological phenomena with respect to the area of interest includes applying, by the computing device, the trained machine-learning model to determine a likelihood that a particular natural resource deposit has formed at or adjacent to the area of interest, and wherein the particular natural resource deposit includes a hydrocarbon reservoir.

11 . The method of claim 7 , wherein applying the trained machine-learning model to the geological data to generate the one or more predictions of geological phenomena with respect to the area of interest includes applying, by the computing device, the trained machine-learning model to determine a paleo-digital elevation model for the area of interest.

12 . The method of claim 7 , wherein the user interface includes:

a background that includes a spatial definition of the area of interest; and

a plurality of indicators, wherein each indicator of the plurality of indicators is positioned on the background in a location on the user interface that corresponds to a corresponding location of the area of interest, and wherein each indicator of the plurality of indicators visually indicates:

a likelihood of a natural resource deposit existing at the corresponding location of the area of interest; or

a relative orogenic age of the corresponding location of the area of interest.

13 . The method of claim 7 , wherein the command includes an indication of whether to form a wellbore in the area of interest.

14 . A non-transitory computer-readable medium comprising instructions that are executable by a processing device for causing the processing device to perform operations comprising:

receiving historical geological data that includes geological data about an area of interest;

pre-processing the historical geological data by applying a selected pre-processing technique to the historical geological data with respect to time-attributed geological phenomena to generate geological data, wherein the selected pre-processing technique adjusts the historical geological data to indicate a temporal relationship between the historical geological data and the time-attributed geological phenomena by:

determining an age of the historical geological data relative to the time-attributed geological phenomena;

generating the geological data by restricting the historical geological data to a time window based on the age of the historical geological data relative to the time-attributed geological phenomena, the time window being before present day; and

splitting the time window into a plurality of bins, each bin of the plurality of bins comprising a binary value that indicates whether a geological phenomenon occurred in a subset of the time window corresponding with the bin;

training a machine-learning model using the geological data to generate a trained machine-learning model;

applying the trained machine-learning model to the geological data to generate one or more predictions of geological phenomena with respect to the area of interest;

providing a user interface that includes a visualization of the one or more predictions of geological phenomena, the visualization including a representation of the area of interest and an indication of a likelihood that the one or more predictions of geological phenomena exists at one or more locations included in the area of interest; and

outputting a command to control a natural resource operation for forming a wellbore in the area of interest, the command comprising instructions that are executable to adjust a direction of a drill bit in response to receiving the one or more predictions of geological phenomena to alter a trajectory of the wellbore.

15 . The non-transitory computer-readable medium of claim 14 , wherein the plurality of pre-processing techniques includes an initial relative timing pre-processing technique, an event barcode pre-processing technique, an event histogram pre-processing technique, a closest age pre-processing technique, and a total count pre-processing technique, and wherein the operation of pre-processing the historical geological data includes:

determining, among a plurality of pre-processing techniques, the selected pre-processing technique to use for pre-processing the historical geological data; and

further pre-processing the historical geological data using a subsequent pre-processing technique of the plurality of pre-processing techniques, the subsequent-pre-processing technique being different than the selected pre-processing technique.

16 . The non-transitory computer-readable medium of claim 14 , wherein the operation of applying the trained machine-learning model to the geological data to generate the one or more predictions of geological phenomena with respect to the area of interest includes applying the trained machine-learning model to determine a likelihood that a particular natural resource deposit has formed at or adjacent to the area of interest, and wherein the particular natural resource deposit includes a hydrocarbon reservoir.

17 . The non-transitory computer-readable medium of claim 14 , wherein the operation of applying the trained machine-learning model to the geological data to generate the one or more predictions of geological phenomena with respect to the area of interest includes applying the trained machine-learning model to determine a paleo-digital elevation model for the area of interest.

18 . The non-transitory computer-readable medium of claim 14 , wherein the user interface includes:

a background that includes a spatial definition of the area of interest; and

a plurality of indicators, wherein each indicator of the plurality of indicators is positioned on the background in a location on the user interface that corresponds to a corresponding location of the area of interest, and wherein each indicator of the plurality of indicators visually indicates:

a likelihood of a natural resource deposit existing at the corresponding location of the area of interest; or

relative orogenic age of the corresponding location of the area of interest.

19 . The system of claim 1 , wherein the operation of training the machine-learning model using the geological data to generate the trained machine-learning model comprises:

inputting the geological data to the machine-learning model;

adjusting, using a first subset of the geological data and based on a loss function, one or more parameters of the machine-learning model to generate the trained machine-learning model; and

validating the trained machine-learning model, wherein validating the trained machine-learning model comprises:

inputting a second subset of the geological data that is different from the first subset into the trained machine-learning model to generate output;

transmitting the output from the trained machine-learning model to a separate model to generate an error metric based on the output and an acceptance criteria; and

validating the trained machine-learning model in response to determining that the error metric matches the acceptance criteria.