Offshore sea surface temperature inversion method based on machine learning

Abstract

The invention discloses an offshore sea surface temperature inversion method based on machine learning, and the method comprises the steps: introducing a sensor visual angle, an initial estimation temperature field, an initial auxiliary temperature field and a time representation variable based on remote sensing image data, and adding combined variable data as feature extension; the defect of insufficient parameter relation expression in a traditional temperature inversion algorithm is overcome; the actual measurement data of the offshore buoy is used as a data set to be input into a data training model, and the influence of mixed pixels on parameter space matching is effectively corrected; and a random forest algorithm is adopted to provide selection indexes of feature importance, an optimal temperature inversion parameter combination is obtained, and a machine learning model is constructed to realize high-precision offshore sea surface temperature inversion under limited actual measurement data.

Description

technical field [0001] The invention relates to the field of satellite remote sensing sea surface temperature inversion, in particular to a sea surface temperature inversion method in offshore sea areas based on machine learning. Background technique [0002] China is one of the largest coastal countries in the world, and there are a large number of mariculture areas and marine ranch demonstration areas in the offshore waters. Sea surface temperature (SST) is an important dynamic factor in coastal waters, and obtaining high-precision sea surface temperature is a necessary condition for monitoring and managing coastal waters. Usually, the traditional sea surface temperature monitoring data is collected through fixed-point ships and buoy measurements, which is difficult to obtain in a timely manner, and the obtained data is small-scale point data. With the development of remote sensing technology, satellite remote sensing has increased the time frequency and spatial coverage ...

AI Technical Summary

Problems solved by technology

[0004] According to the deficiencies in the existing technology, the present invention proposes a machine learning-based offshore sea surface temperature inversion method, which fully combines the measured data of offshore buoys to solve the problem of low inversion accuracy of traditional sea surface temperature inversion methods in offshore waters. Spatial matching is greatly affected by mixed pixels, and the technical problem of insufficient parameter relationship description

Images