# Stochastic uncertainty analysis of gravity gradient tensor components and their combinations

**Authors:** Pejman Shamsipour, Amin Aghaee, Tedd Kourkounakis, Shawn Hood  
**arXiv:** [2205.09159](https://arxiv.org/abs/2205.09159)  
**Date:** 2022-05-18  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2205.09159v1/

## Key points

- Full Tensor Gravity (FTG) devices measure up to five independent components of the gravity gradient tensor
- Lack of quantitative understanding on which tensor components are crucial for accurate subsurface density modeling through gravity inversion
- Researchers proposed a stochastic inversion approach to assess uncertainty associated with tensor components and their combinations
- Geostatistical inversion techniques based on Gaussian process regression and cokriging were applied
- Cokriging variances can indicate the relative significance of different tensor components in recovering a subsurface density model
- The approach was demonstrated using the New Found dataset, showcasing its effectiveness in selecting optimal combinations of tensor components
- This research contributes to advancing FTG data analysis and its potential applications in geophysics
- Provides a quantitative framework for assessing uncertainty and identifying important tensor components

## Summary

In the field of geophysics, Full Tensor Gravity (FTG) devices have been developed to measure up to five independent components of the gravity gradient tensor. However, there is still a lack of quantitative understanding regarding which specific tensor components or combinations of components are most crucial for accurately recovering a subsurface density model through gravity inversion. This knowledge gap exists because different scenarios and purposes may require different components to be prioritized. To address this issue, a team of researchers led by Pejman Shamsipour proposed a stochastic inversion approach to assess the uncertainty associated with gravity gradient tensor components and their combinations. By applying geostatistical inversion techniques based on Gaussian process regression and cokriging, they aimed to quantitatively analyze the importance of different tensor components in various environments. The researchers found that the cokriging variances, which represent the variance function of the Gaussian process, can serve as useful indicators for distinguishing between different gravity gradient tensor components. These variances provide valuable insights into the relative significance of each component in recovering a subsurface density model. To demonstrate the effectiveness of their approach in real-world applications, the team applied it to the New Found dataset. Through this case study, they were able to showcase how their method can aid in selecting optimal combinations of gravity gradient tensor components for accurate subsurface density modeling. Overall, this research contributes significantly to advancing our understanding of FTG data analysis and its potential applications in geophysics. By providing a quantitative framework for assessing uncertainty and identifying important tensor components, this work offers valuable insights for future studies aiming to improve gravity inversion techniques and enhance subsurface imaging capabilities.