SYLVA
A Thermodynamic-Fuel Continuum Framework for Wildfire Spread Rate and Fireline Intensity Estimation
Operational Intelligence System for Mediterranean Wildfire Rapid Spread Forecasting
View on GitLabClosing the Operational Forecasting Gap
Addressing systematic underprediction bias in rapid fire spread forecasting through integrated multi-parameter analysis
Rapid Fire Spread Detection
74% of structure loss and 83% of suppression fatalities are attributable to rapid spread events (≥30 m/min sustained over ≥30 minutes). SYLVA achieves 81-87% accuracy in discriminating rapid spread from non-rapid periods.
Validated Framework
Retrospective validation using 213 Mediterranean wildfire episodes across Greece, Italy, Spain, Portugal, and France from 2,842 field samples and 2000-2024 fire seasons.
Superior Performance
AUC = 0.88, Brier Skill Score = 0.36. Performance exceeds operational fire behavior guidance (BehavePlus, FARSITE) by 14-22% at 60-120 minute lead times.
Real-Time Integration
Integrates automatic weather stations, Sentinel-2 MSI imagery (NDWI inversion), Canadian FFDR System, digital elevation models, and field fuel inventory data.
Thermodynamic Foundation
Treats Mediterranean fuel complexes as coupled thermodynamic-moisture continua where fire spread potential is governed by available heat flux to heat sink capacity ratio.
Mediterranean Calibration
Specifically calibrated for Pinus halepensis, Quercus ilex, Mediterranean maquis, and dry grasslands with regional fuel structure and drought climatology.
The 9 Essential Parameters
Integrated framework synthesizing fuel moisture dynamics, fuel structure, meteorology, and terrain for operational rapid spread assessment
Live Fuel Moisture
Sentinel-2 NDWI inversion at 20m resolution, 5-day revisit. Critical for Mediterranean shrubland fire behavior during summer drought stress.
Dead Fuel Moisture (1-hr)
Fine Fuel Moisture Code (FFMC) from automatic weather stations. Primary driver of surface fire ignition and initial spread in <15% moisture ranges.
Dead Fuel Moisture (10-hr)
Duff Moisture Code (DMC) representing loosely packed organic layers. Influences sustained fire intensity and burn-through depth.
Dead Fuel Moisture (100-hr)
Drought Code (DC) representing deep organic layers. Long-term drought indicator affecting fire season severity and containment difficulty.
Canopy Bulk Density
Landsat-derived with LiDAR validation (0.08-0.45 kg/m³). Critical threshold >0.20 kg/m³ for active crown fire spread in Mediterranean forests.
Surface Fuel Load
Field inventory and photo series (15-65 tons/ha). Combined with fuel bed depth determines available heat release and flame length.
Fuel Bed Depth
Field measurements (0.3-4.0m). Tall shrub fuels (maquis 1.5-4.0m) require modified Rothermel extrapolation beyond laboratory fuel bed calibration.
Terrain-Adjusted Wind Vector
WindNinja/WindWizard algorithms from AWS networks. Topographic acceleration, channeling effects, and aspect-driven circulations in complex terrain.
Vapor Pressure Deficit
Derived from temperature and relative humidity. Drives live fuel moisture loss rates (2-5% per day) during heat wave episodes.
Validation Results
213 Mediterranean wildfire episodes demonstrating operational forecast skill improvements
| Fuel Type | Fire Cases | SYLVA POD | Operational POD | Improvement |
|---|---|---|---|---|
| Pinus halepensis | 68 | 0.86 | 0.71 | +15% |
| Quercus ilex | 42 | 0.81 | 0.67 | +14% |
| Mediterranean maquis | 53 | 0.84 | 0.69 | +15% |
| Dry grassland | 24 | 0.79 | 0.57 | +22% |
Operational Implementation
Integration requirements for civil protection agencies and forest fire management centers
Automatic Weather Stations
10-minute wind speed, air temperature, relative humidity observations. Network density requirements: 1 per 1,000-5,000 km² for operational fuel moisture estimation.
Sentinel-2 MSI Imagery
20m resolution NDWI inversion for live fuel moisture with 5-day revisit interval. Critical for capturing rapid drying during heat waves (2-5% per day moisture loss).
Canadian FFDR System
Fine Fuel Moisture Code (FFMC), Duff Moisture Code (DMC), and Drought Code (DC) for dead fuel moisture timelag dynamics and seasonal drought accumulation.
Digital Elevation Models
Aspect, slope, and topographic wind adjustment algorithms (WindNinja/WindWizard) for terrain-induced acceleration and channeling effects at 20-80m resolution.
Field Fuel Inventory
Fuel load transects, canopy bulk density measurements, canopy base height, and fuel bed depth sampling. Regional calibration campaigns for Mediterranean fuel types.
Open Source Framework
Complete implementation available under CC-BY license. Python-based operational pipeline with real-time data stream integration and probabilistic forecast output.
Access the Framework
Complete SYLVA protocol documentation, validation datasets, and operational implementation code available under Creative Commons Attribution 4.0