Abstract: In the quality control research of remote sensing ocean wave data, traditional quality control methods exhibit certain limitations in detecting single-point outliers due to the inherent complexity and irregularity of wave data. Deep learning, with its powerful feature-learning capabilities and strength in handling nonlinear, demonstrates advantages in addressing nonlinear complex problems. Its application in data quality control can enhance outlier detection performance. In this study, remote sensing significant wave height (SWH) data were used to develop a bi-directional long short-term memory (Bi-LSTM) network model for SWH prediction. A threshold-based approach was then applied to detect outliers, and its performance was compared against 3σ criterion, isolation forest, LSTM, and VAE-LSTM methods. Experimental results demonstrate that the Bi-LSTM–based quality control model achieves superior outlier detection, with precision, recall, F1-score, and runtime of 91%, 93%, 92 and 3.35 s, respectively, indicating its effectiveness for quality control of remote sensing wave data.