3D Visualization

3D Visualization is the space where sampling data, site stratigraphy, and propagation hypotheses are integrated into a single interactive scene. It serves both as a reading tool (viewing a plume, understanding a vertical anomaly) and a workspace (building a subsurface model, simulating a transformation, comparing multiple scenarios). It is the module where everything entered in other parts of IsoFind becomes spatial.

What 3D Visualization allows you to do

The module covers four activities that combine freely within the same project. These activities share the same 3D scene and data, avoiding the typical disconnect between mapping tools, conceptual model tools, and simulation tools.

Configure the Scene

Define the geographical extent, reference altitude, calculation grid, and visual markers.

Build a Subsurface

Stack stratigraphic layers, define their geometry, and set their hydrogeological properties.

Simulate a Transformation

Propagate a contaminant, track a redox reaction, and calculate the temporal evolution of a field.

Compare Hypotheses

Save simulation scenarios and compare their results side-by-side on the same scene.

Interpret and Export

Analyze cross-sections, identify active zones, and produce report blocks associated with the case.

Share the Result

High-definition captures, replayable scenes, and integration into reports and the ISOF archive.

3D Scene Structure

An IsoFind 3D Visualization scene consists of several stacked information layers, each with its own construction and rendering logic. Understanding this stratification helps in knowing where to take action to achieve a specific result.

Level	Content	Source
Geographic Reference	Extent, axes, reference altitude, orientation	Scene Configuration
Subsurface	Stratigraphic layers, lithology, hydrogeological properties	Manual entry or borehole log import
Sampling Points	3D positions of samples (x, y, depth)	samples table
Scalar Fields	Concentrations, isotopic ratios, interpolated physicochemical parameters	sample_molecules, sample_geochem, isotope_data tables
Contamination Sources	Emission points or zones, flow rates, isotopic signatures	Manual entry in the Simulation module
Simulation Results	Fields calculated by the propagation engine	On-the-fly calculation

Rendering Engine

IsoFind uses a local WebGL implementation based on Three.js, embedded within the desktop version. No data is sent over the internet during rendering: the scene is calculated and displayed locally. This architecture meets the sovereignty requirements for sensitive sites and ensures that confidential data never leaves the workstation.

The engine supports six visualization modes for a scalar field, each answering a different question. Mode buttons are grouped in the scene's sidebar.

Mode	Displayed Content	Usage
Points	Raw values at sampling positions	Direct data review without interpolation
IDW Volume	3D volume interpolated via Inverse Distance Weighting	Overview of the observed field
Predicted Prior	Prior calculated by the Nexus bridge (mixing, Rayleigh, speciation)	Geochemical reading without ML correction
Residuals	Discrepancy (δ_observed − δ_prior)	Identify points that the Nexus model alone does not replicate
ML Corrected	Prior + MLP correction trained on residuals	Best global estimate after local learning
Simulation	Result from the 3D ADE plume engine at calculated time steps	Temporal projection and scenarios

The first four modes are available in all editions and only require sampling data. The ML Corrected mode requires a locally trained residual model (dedicated button in the sidebar). The Simulation mode requires source and parameter configuration and produces a field distinct from observations.

For a first look at the module, the recommended workflow is to open a project that already contains samples, switch to Points mode, and then move to IDW Volume mode to see the interpolated field before starting prior calculation and residual ML training.

Accessing the Module

The 3D Visualization module is accessible from the main menu of an open project. The scene is automatically constructed from the data already entered: each sample with geographic coordinates and a depth is positioned, and each associated measurement is available as a scalar field.

Main Menu > Open Project > 3D Visualization > Scene Loaded > Sidebar Tools