Abstract: Compact high-frequency surface wave radar (HFSWR) suffers from low target localization accuracy due to the small aperture of the receiving antenna arrays and low transmission power. This results in significant fluctuations in target parameter sequences, adversely affecting track prediction accuracy and leading to plot-track association errors. To address these challenges, a track prediction method that leverages the spatiotemporal correlation among targets by combining track quality assessment with assistance from target position topology is proposed. First, a real-time track quality evaluation is constructed to identify tracks at risk of crossing or fragmentation, along with their neighboring and reference tracks. Then, a topological structure is built based on the positions of the concerned targets, neighboring targets, and reference targets, and perform prediction and state updates on the reference target tracks. Finally, the updated reference target states and the topological structure are utilized to predict the states of the concerned targets. Experimental results demonstrate that the proposed method effectively improves track prediction accuracy, increases the average track duration by 11%, and resolves issues of mis-tracking and track fragmentation caused by plot-track association errors.