Numerical Simulation of Terrain-Adaptive Wind Field Model Under Complex Terrain Conditions

Complex terrain features such as mountains and hills can obstruct the airflow and force upward motion, thereby altering local atmospheric circulation patterns. During the rainy season, these terrain characteristics are more prone to causing intense local precipitation, leading to geological hazards such as floods and debris flows. These phenomena are closely linked to the intricate influence of terrain on wind fields, highlighting the necessity for in-depth research into wind field characteristics under complex terrain conditions. To address this, we propose a neural-network-based model leveraging terrain data and horizontal wind speed data to predict atmospheric motion characteristics and terrain uplift effects in specific terrain conditions. To enhance the generalization ability of the model, we innovatively extract key physical information from the horizontal wind vector data as training parameters. By comparing with the results of the Fluent model, we validate the model’s capability in dynamic downscaling and flow field modeling. Experimental outcomes demonstrate that our model can generate terrain-adapted convective warning data with a high accuracy, even when terrain features are altered. Under unoptimized conditions, the results at a maximum resolution of 50 m require only 26 s, and the computation time can be further reduced with algorithmic improvements. This research on adaptive wind field modeling under complex terrain conditions holds significant implications for local wind field simulation and severe convective weather forecasting.