This paper investigates a maritime Internet of Things (IoT) network architecture enhanced by a high-altitude platform (HAP) that provides uplink connectivity for shipborne IoT devices (IoTDs). To address the challenges of efficient spectrum utilization and dynamic channel conditions, we employ rate-splitting multiple access (RSMA) to manage the uplink transmission. We formulate a sum-rate maximization problem that jointly optimizes bandwidth allocation, transmit power, and decoding order at the HAP. To address the problem’s non-convexity and high dimensionality, we propose a deep reinforcement learning (DRL) framework based on the deep deterministic policy gradient (DDPG) algorithm. Simulation results demonstrate that the proposed RSMA-enabled system consistently outperforms conventional methods, with notable improvements in spectral efficiency.