To ensure the safe protection of marine engine systems, it is necessary to explore the hot surface ignition (HSI) characteristics of marine diesel in ship environments. However, an accurate model describing these complex characteristics is still not available. In this work, a new experimental method is proposed in order to enhance prediction performance by integrating testing data of the characteristics of HSI of marine diesel. The sensitivity of HSI is determined by various factors such as surface parameters, flow state, and the ship’s environment. According to variations in the HSI status of marine diesel in an engine room, the HSI probability is distributed in three phases. It is essential to determine whether the presence of marine diesel or surrounding items can intensify the risk of an initial fire beginning in the engine room. A vapor plume model was developed to describe the relationship between HSI height and initial specific buoyancy flux in vertical space. Further, field distribution revealed significant variation in the increase in temperature between 200 and 300 mm of vertical height, indicating a region of initial HSI. In addition, increasing surface temperature did not result in a significant change in ignition delay time. After reaching a temperature of 773 K, the ignition delay time remained around 0.48 s, regardless of how much the hot surface temperature increased. This study reveals the HSI evolution of marine diesel in a ship engine room and develops data-based predictive models for evaluating the safety of HSI parameters during initial accident assessments. The results show that the goodness of fit of the predictive models reached above 0.964. On the basis of the predicted results, the HSI characteristics of marine diesel in engine rooms could be gleaned by actively determining the parameters of risk.