Confidence volumes for earth modeling using machine learning

1. A method, comprising:

receiving detected data, wherein the detected data includes formation attributes relating to one or more depth points along or near a wellbore;

providing inputs to a plurality of machine learning models based on the detected data, wherein each machine learning model of the plurality of machine learning models has been trained through an iterative supervised learning process based on labeled training data and a corresponding objective function;

receiving output values from the plurality of machine learning models based on the inputs;

determining a measure of variance among the output values;

generating a confidence indicator related to the output values based on the measure of variance;

omitting a subset of the plurality of machine learning models from an ensemble of machine learning models based on:

the confidence indicator; and

a determination of whether a subset of the output values corresponding to two physical formation attributes satisfy a known relationship between the two physical formation attributes; and

using the ensemble of machine learning models to determine an adjustment to a drilling or reservoir operation, wherein the adjustment to the drilling or reservoir operation comprises a change to: a well placement, a well trajectory, a mud weight, a backpressure, a pump rate, a fluid composition, a casing depth, a weight on bit, rotations per minute, flow rate, a torque on bit, a bit speed, a tripping speed, a rate of penetration, reservoir simulation, artificial lift, water injection, gas injection, enhanced recovery, controlling water production, zonal isolation, or well workover.

2. The method of claim 1 , further comprising selecting a subset of the output values based on the confidence indicator.

3. The method of claim 1 , further comprising displaying the confidence indicator to a user via a user interface.

4. The method of claim 1 , wherein generating the confidence indicator related to the output values based on the measure of variance comprises generating a confidence volume for the wellbore.

5. The method of claim 4 , wherein the confidence volume indicates standard deviations with respect to the output values.

6. The method of claim 4 , wherein the confidence volume is a probabilistic volume.

7. The method of claim 4 , wherein the confidence volume is inputted into a reservoir model.

8. The method of claim 1 , wherein the detected data comprises one or more information types selected from the group consisting of: seismic volumes, seismic geologic maps, seismic images, electromagnetic volumes, checkshots, gravity volumes, horizons, synthetic log data, well logs, mud logs, gas logs, fluid samples, well deviation surveys, isopachs, vertical seismic profiles, microseismic data, drilling dynamics data, initial information from wells, core data, gamma, temperature, torque, differential pressure, standpipe pressure, mud weight, downhole accelerometer data, downhole vibration data, gamma, resistivity, neutron, density, compressional, or shear logs.

9. A system, comprising: one or more processors; and a memory comprising instructions that, when executed by the one or mom processors, cause the system to perform a method, the method comprising:

receiving detected data, wherein the detected data includes formation attributes relating to one or more depth points along or near a wellbore;

providing inputs to a plurality of machine learning models based on the detected data, wherein each machine learning model of the plurality of machine learning models has been rained through an iterative supervised learning process based on labeled training data and a corresponding objective function;

receiving output values from the plurality of machine learning models based on the inputs;

determining a measure of variance among the output values;

generating a confidence indicator related to the output values based on the measure of variance;

omitting a subset of the plurality of machine learning models from an ensemble of machine learning models based on:

the confidence indicator; and

a determination of whether a subset of the output values corresponding to two physical formation attributes satisfy a known relationship between the two physical formation attributes; and

using the ensemble of machine learning models to determine an adjustment to a drilling or reservoir operation, wherein the adjustment to the drilling or reservoir operation comprises a change to: a well placement, a well trajectory, a mud weight, a backpressure, a pump rate, a fluid composition, a casing depth, a weight on bit, rotations per minute, flow rate, a torque on bit, a bit speed, a tripping speed, a rate of penetration, reservoir simulation, artificial lift, water injection, gas injection, enhanced recovery, controlling water production, zonal isolation, or well workover.

10. The system of claim 9 , wherein the method further comprises selecting a subset of the output values based on the confidence indicator.

11. The system of claim 9 , wherein the method further comprises displaying the confidence indicator to a user via a user interface.

12. The system of claim 9 , wherein generating the confidence indicator related to the output values based on the measure of variance comprises generating a confidence volume for the wellbore.

13. The system of claim 12 , wherein the confidence volume indicates standard deviations with respect to the output values.

14. The system of claim 12 , wherein the confidence volume is a probabilistic volume.

15. A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors of a computing system, cause the computing system to perform a method, the method comprising:

receiving detected data, wherein the detected data includes formation attributes relating to one or more depth points along or near a wellbore;

providing inputs to a plurality of machine learning models based on the detected data, wherein each machine learning model of the plurality of machine teaming models has been trained through an iterative supervised learning process based on labeled training data and a corresponding objective function;

receiving output values from the plurality of machine learning models based on the inputs;

determining a measure of variance among the output values;

generating a confidence indicator related to the output values based on the measure of variance;

omitting a subset of the plurality of machine learning models from an ensemble of machine learning models based on:

the confidence indicator; and

a determination of whether a subset of the output values corresponding to two physical formation attributes satisfy a known relationship between the two physical formation attributes; and

using the ensemble of machine learning models to determine an adjustment to a drilling or reservoir operation, wherein the adjustment to the drilling or reservoir operation comprises a change to: a well placement, a well trajectory, a mud weight, a backpressure, a pump rate, a fluid composition, a casing depth, a weight on bit, rotations per minute, flow rate, a torque on bit, a bit speed, a tripping speed, a rate of penetration, reservoir simulation, artificial lift, water injection, as injection, enhanced recovery, controlling water production, zonal isolation, or well workover.