Data and knowledge combined driven interpretable sea temperature prediction method and device

The invention discloses a data and knowledge combined driven interpretable sea temperature prediction method and device, and the method comprises the steps: extracting spatial global and local features of sea surface temperature by employing ASPP, extracting frequency domain features by employing Fourier transform, and splicing the frequency domain features with the spatial global and local features to form space-time fusion features of the sea surface temperature; coupling modeling of radiation flux data is carried out according to the ocean surface heat balance equation, the interdependence relation between the flux data is obtained, feature fusion is achieved through 3D-CNN, and multivariable coupling features of radiation flux are formed; pINN is used as a prediction model, channel splicing is carried out on the space-time fusion feature of the sea surface temperature, the multivariable coupling feature of the radiation flux and the ocean kinetic parameters to form a multi-modal feature, the multi-modal f