Spatiotemporal Big Data Retrieval Method Based on GeoHash Algorithm and Multimode Composite Index

With the increasing demand for multisource heterogeneous spatiotemporal data in applications such as smart cities and intelligent transportation, how to balance retrieval accuracy, multimodal adaptation, and high concurrency response has become a key issue. This study proposes a spatiotemporal big data retrieval model that integrates a geographic hash coding algorithm and multimode combination index optimization. The model first uses geographic hashing combined with row key and nonrow key indexes to filter candidate sets. Then, it combines spatial, temporal, and multimodal feature indexes to complete cross‐dimensional matching and rearrangement. It also introduces a skip table structure and parameter adaptive strategy to optimize latency and concurrency performance. The experiment showed that the model had the highest precision of 94.27% on public datasets, the highest recall rate of 93.05%, F 1 values exceeding 93%, and the lowest retrieval delay controlled at 121.49 ms, significantly better than other methods. The coverage rates in urban and rural nodes were 91.2% and 88.7%, with the lowest redundancy rate of only 12.4% under vehicle and pedestrian interference. The results showed that the model had advantages in precision, latency, and energy efficiency and is suitable for multiscenario applications such as smart cities, intelligent transportation, and emergency management. This study can provide an effective technical support for the future development of big data retrieval technology.