This research proposes a simulation dataset that reflects the operational characteristics of submarine infrared sensors in maritime environments. Although infrared imaging is widely used for covert observation and surveillance, image quality often degrades due to transmittance attenuation caused by submarine sapphire optical windows and the inherent limitations of Noise Equivalent Temperature Difference (NETD) characteristics. To address this, a realistic dataset was constructed using Radiation Transfer Modeling (RTM), incorporating sapphire optical window transmittance values (0.65 and 0.89) and simulating thermal noise across various NETD levels (10, 20, and 30 mK). Using the generated dataset, noise reduction and super-resolution experiments were performed to evaluate the limitations of existing image enhancement techniques. The proposed dataset offers a standardized benchmark for infrared image enhancement research in specialized underwater environments and enables more accurate prediction of image processing performance under real submarine operating conditions.