1- Overview of the 3D Visualization System
This system was developed to visualize numerical weather prediction (NWP) data from the Korea Meteorological Administration in a web-based 3D environment. The primary goal is to support meteorologists in their analysis of weather phenomena by rendering NWP data alongside topographical information and various weather elements. For 3D visualization, the system employs CesiumJS, a specialized open-source engine for geospatial data rendering. CesiumJS enables high-performance visualization of terrain, spatial models, and time-based data directly in a browser, providing an optimal foundation for the system.

2 - Processing Workflow of NWP Model Data
The system includes an automated pipeline that handles the collection, transformation, and management of large-scale NWP data. This allows for fast and efficient visualization of real-time or periodically updated datasets. Manual reprocessing is also supported, ensuring adaptability and control when necessary.

The processed data is categorized into two main types:

• Isosurface -> GLTF

• Streamline -> JSON

1. Isosurface Generation
   To extract isosurfaces from volumetric scalar fields, the marching cubes algorithm was applied. This method generates triangle mesh representations of 3D structures corresponding to specific threshold values in the data. The generated structures are then converted into GLTF format. Geospatial coordinates are embedded into these GLTF models to ensure accurate placement in the 3D environment.

The models undergo a lightweight optimization process to reduce file size while preserving visual fidelity. This makes them suitable for smooth rendering in web browsers without compromising detail.

1. Streamline Visualization
   Streamlines are used to represent directional flow patterns such as wind. For this, a computed engine calculates flow data in real time on a per-frame basis, and the results are merged into a single texture. This process is executed through parallel computation, ensuring optimized performance even when handling large datasets.

This approach allows for dynamic visualization responsive to user interaction and effectively conveys directional elements such as wind speed and direction.

3 - Significance and Presentation Goal
The described techniques offer a visual aid in interpreting complex numerical weather data, helping meteorologists and researchers understand and utilize this information more intuitively. The system is under continuous development, with additional NWP models being processed and integrated.

This presentation aims to showcase the data processing pipeline and visualization results of the system, introducing new possibilities in weather data interpretation and the technical approaches used to achieve it.