EGU24-2442, updated on 08 Mar 2024
https://doi.org/10.5194/egusphere-egu24-2442
EGU General Assembly 2024
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.

GAIN, a Machine Learning approach for Airborne, Maritime, and Submarine Gravimeter Systems

Lorenzo Iafolla1, Massimo Chiappini1, and Francesco Santoli2
Lorenzo Iafolla et al.
  • 1Istituto Nazionale di Geofisica e Vulcanologia, Ambiente, Rome, Italy (lorenzo.iafolla@ingv.it)
  • 2Istituto Nazionale di Astrofisica, IAPS, Rome, Italy

Precision gravimeters deployed onboard aerial, groundbased, and underwater moving platforms face significant accuracy challenges due to environmental disturbances such as non-inertial reference systems and temperature variations. The “Gravimetro Aereo INtelligente” (GAIN) concept represents a step forward in tackling this problem by using a data post-processing approach. This approach avoids cumbersome, heavy, and power-intensive active compensation systems, thus increasing the instrument's adaptability to small moving platforms.

The GAIN concept is based on three pillars that define its approach. Firstly, it incorporates a multi-sensor system within the gravimeter framework, which might include a three-axial accelerometer, a three-axial gyroscope, multiple thermometers and a barometer. This set of sensors are designed to measure both the effects of gravity and of other disturbances. By utilizing this information, GAIN employs machine learning algorithms (the second pillar) to map the complex relationship between the measurements and the desired gravity value. However, machine learning heavily relies on the availability of high-quality training datasets, which are often scarce and challenging to obtain in operational environments. To address this bottleneck, the third pillar of GAIN utilizes a training platform that can simulate a wide range of environmental situations in a controlled laboratory setting. This platform enables the generation of labeled data that mimics real-world operational scenarios.

This contribution will present the details of the initial GAIN experimental setup, highlighting the successful integration of a multi-sensor system with the training platform. Additionally, early findings will be shared, demonstrating the potential of the GAIN technique in mitigating temperature changes in gravimeters. Finally, the progress of ongoing experiments will be showcased, as we work towards expanding the capabilities of the GAIN method to also address rotations and linear accelerations as sources of interference.

How to cite: Iafolla, L., Chiappini, M., and Santoli, F.: GAIN, a Machine Learning approach for Airborne, Maritime, and Submarine Gravimeter Systems, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-2442, https://doi.org/10.5194/egusphere-egu24-2442, 2024.

Comments on the supplementary material

AC: Author Comment | CC: Community Comment | Report abuse

supplementary materials version 1 – uploaded on 07 May 2024, no comments