Internal waves are widely distributed in the ocean, and their amplitude can reach hundreds of meters and can affect the marine environment.
Remote sensing is an important method for observing internal waves in the ocean. However, inversion of the internal wave amplitude from the remote sensing image is not available.
Recently, a research team led by Dr. LIXiaofeng of the Institute of Oceanography of the Chinese Academy of Sciences (IOCAS) applied. artificial intelligence (AI) Transfer learning techniques for integrating data from laboratories, buoys, and remote sensing to study internal waves.
The study was published in Remote sensing of the environment February 9th.
Researchers apply on the spot, Remote sensing Data to establish a matching dataset Model trainingUsing AI transfer learning technology, we solved various data source problems and accurately reconstructed the 3D structure of internal waves.
The AI transmission learning-based internal isolated wave amplitude inversion model is a two-step model.In the first stage, Transfer learning is applied to handle various internal wave data sources. A coordinated change inspired by ResNet called a short connection has been introduced. In the second stage, the model uses the actual ocean density information to perform the bias correction of the first stage results.
“The model can invert the amplitude of the internal wave with the information extracted from the satellite image as input. Internal wave It can be rebuilt. “
Related results show that it is reliable and feasible to develop inversion models of complex oceanic phenomena based on pure data-driven ocean information big data.
“As a fast-growing emerging technology, AI technology can establish fast, direct mapping relationships when studying complex marine phenomena,” says Dr. LI.
Xudong Zhang et al, Ocean Internal Wave Amplitude Search from Satellite Images Based on Data-Driven Transmission Learning Model, Remote sensing of the environment (2022). DOI: 10.1016 / j.rse.2022.112940
Chinese Academy of Sciences
Quote: AI transfer learning technology is an ocean internal wave acquired from https://phys.org/news/2022-02-ai-techniques-ocean-internal.html on February 17, 2022 (February 17, 2022). Useful for day) investigations
This document is subject to copyright. No part may be reproduced without written permission, except for fair transactions for personal investigation or research purposes. Content is provided for informational purposes only.
AI transfer learning technology is useful for studying ocean internal waves
Source link AI transfer learning technology is useful for studying ocean internal waves