Volume Rendering Techniques in 3D GIS: Triangulation Mesh, Voxel Grid, and Volume Rendering

3D visualization engines are a crucial part of Tongtu GIS, requiring not only high visual quality and special effects but also extensive research on spatial analysis, performance optimisation, and data organisation; among many 3D techniques, volume‑display methods are especially important for geology, meteorology, oceanography, land‑planning and city‑management applications.

Triangulation Mesh Scheme – In WebGL the basic primitive is the triangle, which consists of three vertices each holding position, colour and other attributes. A surface model is built from many seamless triangles, forming the model’s skeleton and skin. This approach quickly produces a visual product from a small amount of data, making it suitable for surface‑oriented scenarios such as mountain or terrain modelling. However, its simplicity limits the ability to expose internal information.

Understanding WebGL basics, a triangle is composed of three vertices, each containing position, colour and other attribute information. A surface model is essentially a collection of seamless triangles, whose combination determines the model’s "skeleton" and "skin".

Advantages: fast production with minimal data. Disadvantages: cannot reveal internal model details.

Examples: (1) Triangulated mountain model with lighting; (2) Smoothed underground rock‑layer model.

Voxel‑Grid Scheme – A voxel grid extends the concept of raster data into three dimensions, representing each cell (voxel) with an intensity value. Building a voxel model requires spatial interpolation to generate a regular 3‑D grid from irregular source points. The process includes constructing a cubic grid based on point distribution and interpolating attribute values for each grid node. If the source data is already a regular grid, interpolation is unnecessary.

Raster data is a basic GIS model that uses a regular array of cells to represent spatial phenomena; voxel data is the three‑dimensional counterpart, using voxels to express intensity values.

Advantages: rich internal attribute information enables layer‑by‑layer analysis, filtering and filling operations. Disadvantages: large preprocessing workload due to grid construction and interpolation.

Suitable scenarios: sunlight analysis, atmospheric pollution analysis, signal‑strength attenuation, etc.

Instanced Rendering Optimization – Rendering each small cube individually is costly; instanced drawing creates a single draw call that renders many instances of the same geometry, greatly reducing memory and GPU overhead.

Instanced rendering can be thought of as "stamping" many copies of a model using a single draw command, similar to using a stamp to imprint multiple marks.

The Tongtu GIS platform has implemented instanced rendering for various geometric models, including small cubes.

Volume Rendering Scheme – Volume rendering visualises internal data by sampling voxels along view‑rays, accumulating colour and intensity to produce a final pixel colour. Although WebGL lacks native 3‑D textures, the technique can be approximated by packing 3‑D slices into a 2‑D texture and mapping 3‑D coordinates to 2‑D texture coordinates during ray sampling.

The implementation can borrow from ray‑casting: a ray enters the volume at P1, exits at P2, and the colour of P1 is determined by accumulating contributions from all voxels the ray traverses, performed in the fragment shader.

Advantages: spectacular visual effect that clearly reveals internal structures. Disadvantages: complex pipeline requiring voxel pre‑computation, texture stitching, and ray‑casting logic.

Suitable for 3‑D distribution maps, scans, fire, smoke, clouds, atmospheric or temperature fields.

In summary, the article presented three volume‑display methods—triangulation mesh, voxel‑grid, and volume rendering—described their technical foundations, pros and cons, and appropriate application scenarios. Selecting the proper technique depends on business requirements, and future work will continue to optimise and expand these visualisation capabilities for various industries.