Traditional lane detection algorithms often struggle in challenging conditions, such as adverse weather or poor visibility, making them prone to missed or incorrect detections. Recent approaches have attempted to incorporate High Definition (HD) maps or in-vehicle sensors as prior information. Due to the inherent limitations of in-vehicle sensors and the difficulty of obtaining HD maps significantly degrade the performance of these methods in complex scenarios. In this study, we explore a novel perspective by leveraging satellite maps and Standard Definition (SD) maps to complement in-vehicle sensors. Additionally, we design a polyline-based representation method and a hierarchical fusion module, which significantly and consistently enhance the performance of state-of-the-art online mapping methods in both lane detection and topology.