As marine operations advance into ultra-deep waters, the demand for innovative technologies and intelligent equipment continues to grow to ensure operational safety under increasingly harsh environmental conditions [1]. The operational window of vessels is highly dependent on weather conditions, necessitating an accurate and timely understanding of the marine environment to reduce operational costs and enhance navigational safety [2]. In recent years, there has been a discernible shift toward the development of intelligent and autonomous vessels capable of performing varying levels of decision-making and control in complex maritime environments. In this context, constructing a real-time and robust sea state estimation model has become essential for supporting the situational awareness and decision-making of autonomous ships.