Calibration of CFOSAT Off-Nadir SWIM SWH Product Based on PSO-BPNN Model
High-precision significant wave height(SWH)observation is crucial for Marine research.The Surface Waves Investigation and Monitoring(SWIM)aboard the China-France Oceanic Satellite(CFOSAT)provides the global wave data.However,the SWH data obtained from SWIM off-nadir measurements exhibit a relatively high bias when compared with true value,thereby demonstrating the need for calibration.In this paper,the Back Propagation Neural Network(BP)based on Particle Swarm Optimization(PSO)is established,and the wind field data provided by the wind Sactterometer(SCAT)sensor carried by the CFOSAT is used to correct the SWH by the SWIM 6°off-nadir.The applicability of the model to correct the SWH of 8°and 10°was verified.The proposed model has been compared with various machine learning models,and validated across different sea conditions.The results indicate that the root mean square error of the corrected SWH is 0.232 9 m,with a correlation coefficient(R)of 0.985 8,and that the SWH at 8° and 10° can also be corrected with high precision.The model has the highest accuracy under medium sea state,but there are overestimation and underestimation in low and high state.This bias can be reduced by increasing the training data of low and high sea state.Finally,comparing different machine leaning models,it is proved that the PSO optimization algorithm can improve the accuracy of the model.The PSO-BPNN model proposed improves the accuracy of the SWH parameters,and provides an experimental idea for the study of parameter correction.
万方数据
维普
