Identifying hydrocarbon reserves of a subterranean region using a reservoir earth model that models characteristics of the region

1. A computer-implemented method for determining characteristics of an underground formation in a subterranean region of a geological area, the method comprising:

obtaining a first wavefield represented by seismic data generated from a plurality of sampling sensors, wherein a subset of the sampling sensors are deployed in the subterranean region;

providing data values of the seismic data that indicate properties of the underground formation as inputs to a machine-learning engine configured to generate one or more models;

processing the inputs corresponding to data values of the seismic data using the machine-learning engine;

in response to processing the data values of the seismic data, generating a plurality of predictive models, each predictive model comprising a neural network having interconnected nodes that are configured to determine geological properties of a layer in the underground formation based on a respective analytical rule of the predictive model;

providing, to each of the predictive models, new data values of seismic data representing a second wavefield obtained using the subset of sampling sensors;

automatically updating the respective analytical rule of each predictive model in response to processing the new data values of seismic data at the predictive model; and

determining, from the new data values of seismic data, (i) a first geological property of the layer using the updated analytical rule of a first predictive model and (ii) a second, different geological property of the layer using the updated analytical rule of a second, different predictive model.

2. The method of claim 1 , comprising:

generating an integrated multi-dimensional geological model based on the plurality of predictive models, wherein the integrated multi-dimensional geological model is configured to model characteristics of reservoirs in the subterranean region to estimate hydrocarbon reserves using at least the first and second geological properties of the layer in the underground formation.

3. The method of claim 2 , wherein obtaining each of the first and second wavefields comprises:

obtaining each of the first and second wavefields in response to drilling the subterranean region to penetrate one or more layers in the underground formation.

4. The method of claim 3 , comprising:

determining, by the integrated multi-dimensional geological model, a position of one or more well bores in the subterranean region based on the modeled characteristics of reservoirs in the subterranean region and estimates of hydrocarbon reserves in the reservoirs.

5. The method of claim 3 , comprising:

determining respective quality measures of sediments in each of the one or more layers using each predictive model of the plurality of predictive models; and

based on the respective quality measure of sediments in each of the one or more layers, determining, by the integrated multi-dimensional geological model, a trajectory for drilling the subterranean region to penetrate the one or more layers in the underground formation.

6. The method of claim 1 , wherein generating the plurality of predictive models comprises:

generating a three-dimensional geological numerical model configured to predict numerical values indicating one or more properties of the layer in the underground formation.

7. The method of claim 6 , wherein generating the plurality of predictive models comprises generating a plurality of permanently active autonomous predictive models.

8. The method of claim 1 , wherein processing the inputs corresponding to the data values of the seismic data comprises:

processing the inputs using one or more neural networks of the machine-learning engine based on analytical rules executed at the machine-learning engine, wherein each of the one or more neural network is configured to represent a respective data model of the machine-learning engine.

9. The method of claim 8 , wherein:

at least one of the analytical rules is a deep-learning algorithm that is executed to process the inputs through one or more layers of a neural network; and

the neural network is implemented on a hardware circuit accessible by the machine-learning engine.

10. The method of claim 1 ,

wherein at least one predictive model is configured to determine formation tops of the layer and at least one predictive model is configured to determine a geological property of the layer other than the top of the layer;

the method further comprising generating an integrated multi-dimensional geological model based on the plurality of predictive models, wherein the integrated multi-dimensional geological model is configured to determine suitable locations for deviated wells in the formation.

11. A system for determining characteristics of an underground formation in a subterranean region of a geological area, the system comprising:

one or more processing devices and one or more non-transitory machine-readable storage devices storing instructions that are executable by the one or more processing devices to cause performance of operations comprising:

obtaining a first wavefield represented by seismic data generated from a plurality of sampling sensors, wherein a subset of the sampling sensors are deployed in the subterranean region;

providing data values of the seismic data that indicate properties of the underground formation as inputs to a machine-learning engine configured to generate one or more models;

processing the inputs corresponding to data values of the seismic data using the machine-learning engine;

in response to processing the data values of the seismic data, generating a plurality of predictive models, each predictive model comprising a neural network having interconnected nodes that are configured to determine geological properties of a layer in the underground formation based on a respective analytical rule of the predictive model;

providing, to each of the predictive models, new data values of seismic data representing a second wavefield obtained using the subset of sampling sensors;

automatically updating the respective analytical rule of each predictive model in response to processing the new data values of seismic data at the predictive model; and

determining, from the new data values of seismic data, (i) a first geological property of the layer using the updated analytical rule of a first predictive model and (ii) a second, different geological property of the layer using the updated analytical rule of a second, different predictive model.

12. The system of claim 11 , wherein the operations comprise:

generating an integrated multi-dimensional geological model based on the plurality of predictive models, wherein the integrated multi-dimensional geological model is configured to model characteristics of reservoirs in the subterranean region to estimate hydrocarbon reserves using at least the first and second geological properties of the layer in the underground formation.

13. The system of claim 12 , wherein obtaining each of the first and second wavefields comprises:

obtaining each of the first and second wavefields in response to drilling the subterranean region to penetrate one or more layers in the underground formation.

14. The system of claim 13 , wherein the operations comprise:

determining, by the integrated multi-dimensional geological model, a position of one or more well bores in the subterranean region based on the modeled characteristics of reservoirs in the subterranean region and estimates of hydrocarbon reserves in the reservoirs.

15. The system of claim 13 , wherein the operations comprise:

determining respective quality measures of sediments in each of the one or more layers using each predictive model of the plurality of predictive models; and

based on the respective quality measure of sediments in each of the one or more layers, determining, by the integrated multi-dimensional geological model, a trajectory for drilling the subterranean region to penetrate the one or more layers in the underground formation.

16. The system of claim 11 , wherein generating the plurality of predictive models comprises:

generating a three-dimensional geological numerical model configured to predict numerical values indicating one or more properties of the layer in the underground formation.

17. The system of claim 16 , wherein generating the plurality of predictive models comprises generating a plurality of permanently active autonomous predictive models.

18. The system of claim 11 , wherein processing the inputs corresponding to the data values of the seismic data comprises:

processing the inputs using one or more neural networks of the machine-learning engine based on analytical rules executed at the machine-learning engine, wherein each of the one or more neural network is configured to represent a respective data model of the machine-learning engine.

19. The system of claim 18 , wherein:

at least one of the analytical rules is a deep-learning algorithm that is executed to process the inputs through one or more layers of a neural network; and

the neural network is implemented on a hardware circuit accessible by the machine-learning engine.

20. The system of claim 11 , wherein at least one predictive model is configured to determine formation tops of the layer and at least one predictive model is configured to determine a geological property of the layer other than the top of the layer, the operations further comprising:

generating an integrated multi-dimensional geological model based on the plurality of predictive models, wherein the integrated multi-dimensional geological model is configured to determine suitable locations for deviated wells in the formation.