The Construction of a Multimodal Knowledge Graph for WRF Simulation Knowledge

The development of open geographical simulation holds significant importance in advancing geographical research. With its rapid progress, studies on geographical data sharing, open‐source virtual models, and computing resource adaptation have become increasingly mature. However, the sharing of simulation knowledge remains a core bottleneck that urgently needs to be addressed. As an important branch of geographical simulation, the Weather Research and Forecasting (WRF) model's simulation knowledge is characterized by its spatiotemporal and multimodal features, posing challenges to efficient sharing. To tackle this issue, this study proposes a multimodal knowledge graph construction framework based on BiLSTM‐CRF, cosine similarity, and PDFFigures 2.0, constructs a multimodal knowledge graph for WRF simulation knowledge, and develops a retrieval platform. User feedback indicates that this research has facilitated the transition from data services to knowledge services, enhanced the sharing and reuse of WRF simulation knowledge, and provided guidance for simulation experiments and decision‐making.