Reducing effect of motion on nmr measurements

A non-transitory computer-readable storage medium comprising instructions for removing motion-induced noise that, when loaded into a processor, cause the processor to execute steps of

• receiving a plurality of datasets from a nuclear magnetic resonance (NMR) tool, each dataset associated with a respective depth in a wellbore;
• generating an echo train from each dataset;
• identifying one or more datasets that were collected while a motion of the tool was below a predetermined threshold as stationary datasets, wherein datasets not identified as stationary are identified as moving datasets;
• building a statistical model of the echo train;
• estimating a motion-induced noise component of the echo train using the statistical model and the stationary datasets;
• calculating a distortion of the echo train for each datasets using a physical model of the tool within the wellbore;
• estimating a true signal of each moving dataset based on the estimated motion-induced noise component and the respective calculated distortion;
• estimating a true signal of each stationary dataset based on the respective calculated distortion;
• and
• combining the estimated true signals of the moving datasets and the stationary datasets to produce a superset of clean NMR data.

A system for removing motion-induced noise, comprising a processor and a non-transitory computer-readable storage medium coupled to the processor and comprising instructions that, when loaded into the processor, cause the processor to execute steps of:

• receiving a plurality of datasets from a nuclear magnetic resonance (NMR) tool, each dataset associated with a respective depth in a wellbore;
• generating an echo train from each dataset;
• identifying one or more datasets that were collected while a motion of the tool was below a predetermined threshold as stationary datasets, wherein datasets not identified as stationary are identified as moving datasets;
• building a statistical model of the echo train;
• estimating a motion-induced noise component of the echo train using the statistical model and the stationary datasets;
• calculating a distortion of the echo train for each datasets using a physical model of the tool within the wellbore;
• estimating a true signal of each moving dataset based on the estimated motion-induced noise component and the respective calculated distortion;
• estimating a true signal of each stationary dataset based on the respective calculated distortion;
• and
• combining the estimated true signals of the moving datasets and the stationary datasets to produce a superset of clean NMR data.