Predicting Model of Sea Surface Temperature Based on Improved Variational Mode Decomposition and Least Square Support Vector Machine

Accurate prediction of sea surface temperature is very important for marine scientific research and application. Aiming at the nonlinear characteristics of sea surface temperature series, combined predicting model of sea surface temperature based on improved variational mode decomposition and least square support vector machine is proposed. In order to solve the problem that the decomposition level and penalty coefficient of variational mode decomposition (VMD) need to be set manually, dream optimization algorithm (DOA) is introduced to optimize its parameters, and then improved VMD based on DOA is proposed, called DVMD. Aiming at problem that the penalty coefficient and kernel parameters of least squares support vector machine (LSSVM) are randomly selected, LSSVM based on DOA is proposed, called DOA-LSSVM. Firstly, DVMD decompose the sea surface temperature data, and obtain some intrinsic mode functions (IMFs). Secondly, DOA-LSSVM is used to predict each IMFs component. Finally, superimpose and reconstruct the prediction results of different components to obtain the final sea surface temperature prediction value. The sea surface temperature data from January to February, 2022 in National Ocean Science Data Center are selected, and the proposed model is compared with five comparison models. The results show that the RMSE, MAE, MAPE and $\mathrm{R}^{2}$ of the proposed model are 0.5002, 0.4021,1.8032 and 0.9872, respectively, and all the indicators are better than comparison model. The research provides a new method for sea surface temperature prediction.