Abstract
This thesis presents the design and implementation of a real-time, interactive three-dimensional geospatial visualization platform for wildfire monitoring and analysis. The system integrates satellite-derived fire detection data from the NASA Fire Information for Resource Management System (FIRMS) API with a photorealistic digital twin of the Earth’s surface rendered inside the Unity game engine, enhanced by Cesium for Unity, which provides streaming global terrain, satellite imagery, and georeferenced coordinate systems at planetary scale.
The platform allows users to observe historical fire events over a user-defined date range, animate fire spread through a day-by-day timeline, visualize burn scars on the terrain, overlay spatial wind and temperature data, and navigate the three-dimensional scene with a free-fly camera. All data is sourced from open, publicly available APIs and is processed through an automated pipeline that runs entirely within the Unity Editor environment, requiring no external GIS software.
The core technical challenge addressed is bridging the gap between geospatial coordinate systems used in Earth observation data (WGS84 latitude/longitude) and the local Cartesian coordinate system used by the Unity renderer. Cesium for Unity provides the mathematical and runtime infrastructure to resolve this gap, enabling objects placed at geographic coordinates to appear correctly on a photorealistic 3D globe in real-time.
Thesis Details
Key Features
- Automated data pipeline — Fetches and processes NASA FIRMS fire detection records directly inside the Unity Editor; no external GIS software required.
- Three boundary modes — Named preset regions, user-defined custom bounding box, or imported shapefile boundary for flexible study area selection.
- Day-by-day timeline animation — Fires are grouped by detection date and animated chronologically to show the temporal progression of wildfire events.
- Fire clustering and LOD — Nearby fires are merged into scaled representative markers to maintain interactive frame rates with tens of thousands of detections.
- Burn scar visualization — Accumulated burn areas are projected onto the Cesium terrain surface and updated each day of the timeline.
- Weather data overlays — Temperature, wind vectors, and cloud cover layers loaded from Open-Meteo JSON data and rendered on the minimap.
- Orthographic minimap — A top-down 2D minimap displays fire positions, weather layers, and scene orientation in real time.
- Fly-through camera with annotations — Scripted camera paths with timed annotation cards allow presentation-quality scene walkthroughs and video export (MP4 / GIF).
- Confidence filtering — VIIRS low/nominal/high confidence classes are normalized to numeric scores; fire points below a configurable threshold are excluded from visualization.
- Performance presets — Quality and Performance runtime presets adjust particle budgets, smoke LOD, wind grid stride, and render scale from the in-game settings panel.
Study Areas
Four geographically and ecologically diverse study areas were selected to validate the platform’s adaptability across different fire regimes and terrain complexities.
| Region | Country | Area (km²) | Fire Regime | Key Challenge |
|---|---|---|---|---|
| Gran Chaco | Paraguay | ~719,000 | Savanna / agricultural | Baseline; highest detection density (13,279 fires in test range) |
| Iberian Peninsula | Portugal | ~118,000 | Mediterranean chaparral / eucalyptus | Steep terrain; high FRP crown fires; terrain snapping accuracy |
| Eastern Australia (NSW/QLD) | Australia | ~980,000 | Eucalyptus forest / temperate woodland | Large-scale load; 2019–20 Black Summer; clustering performance |
| Hawaiian Islands | USA (Hawaii) | ~28,000 | Tropical dry forest / grassland | Volcanic terrain extremes; 2023 Lahaina fire; sparse distribution |
System Workflow Overview
- Boundary definition — User selects a preset region, draws a custom bounding box, or imports a shapefile boundary in the Unity Editor pipeline tool.
- FIRMS data fetch — The pipeline queries the NASA FIRMS Area API, automatically chunking large date ranges into 5-day windows and merging/deduplicating results.
- Shapefile export — Fire point coordinates are exported as a shapefile (with WGS84 .prj) and a CSV report for archival and GIS compatibility.
- Scene initialisation — Unity loads the shapefile, applies confidence filtering, and waits for Cesium terrain tiles to stream before spawning fire objects.
- Coordinate transformation — Each fire’s WGS84 latitude/longitude is converted to Unity world space via
CesiumGeoreferenceECEF transformation. - Timeline construction — Fire records are grouped by date, clustered for LOD, and stored; timeline autoplay advances day markers and burn scars incrementally.
- Interactive exploration — User navigates in real time: toggles weather overlays, adjusts settings, records fly-through paths, and exports video.