Ocean internal wave forecasting method based on machine learning and remote sensing data

Abstract

The invention provides an ocean internal wave forecasting method based on machine learning and remote sensing data, and the method comprises the steps: firstly obtaining remote sensing image data containing internal wave features, carrying out the preprocessing, obtaining the spatial information and time information of an internal wave through extraction, and constructing an internal wave quasi-synchronous remote sensing image sample library; based on the internal wave quasi-synchronous remote sensing image sample library, constructing an internal wave propagation forecasting model by using multi-dimensional information fusion of a convolutional neural network and a convolutional long-term and short-term memory network to establish an ocean internal wave forecasting model, wherein the input of the forecasting model is environmental factors including water depth, ocean stratification, density difference and internal wave factors, and comprises an internal wave scale and initial wave crest line position data of the internal wave; and the output of the forecasting model is forecasting crest line position data of the internal wave. According to the method, internal wave remote sensingbig data mining is carried out by utilizing strong nonlinear mapping capability and multi-modal fusion capability of machine learning, and timely and accurate prediction of internal waves is realized.

Description

technical field [0001] The invention belongs to the technical field of ocean observation, and in particular relates to an ocean internal wave prediction method based on machine learning and remote sensing data. Background technique [0002] Internal waves are generated and propagated inside the ocean, with a maximum amplitude of 240m and a crest line of hundreds of kilometers, which can propagate hundreds of thousands of kilometers in the ocean and are widely distributed in the global ocean. The propagation of internal waves has an important impact on marine engineering, ecology and military affairs, and has become a hot issue for scholars to study. Remote sensing observation has the advantages of all-day, all-weather, and large-scale width, so multi-source remote sensing observation has become an important means of internal wave research. Some scholars have used remote sensing images to carry out related research on the temporal and spatial distribution, generation, propag...

AI Technical Summary

Problems solved by technology

[0004] Therefore, based on the existing model method, it is impossible to carry out long-term, timely and accurate propagation prediction of internal waves.

Images