Estimation of Prometheus fuel types using physically based remote sensing techniques

[EN] Background Detailed knowledge of the spatial distribution of vegetation fuels is essential for assessing wildfire hazard and behavior, as well as for planning effective management. In southern Europe, the Prometheus project has proposed the differentiation of seven fuel types, but their charact...

Descripción completa

Detalles Bibliográficos
Autores: Fernández Guisuraga, José Manuel, Monzón González, Andrea, Fernández García, Víctor, Peña Pérez, Sergio Alberto, Calvo Galván, María Leonor
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2025
País:España
Institución:Universidad de León
Repositorio:BULERIA. Repositorio Institucional de la Universidad de León
OAI Identifier:oai:buleria.unileon.es:10612/24965
Acceso en línea:https://fireecology.springeropen.com/articles/10.1186/s42408-025-00373-4
https://hdl.handle.net/10612/24965
Access Level:acceso abierto
Palabra clave:Ecología. Medio ambiente
Ingeniería forestal
LiDAR
Mediterranean Basin
MESMA
Random forest
SAR
Sentinel‑2
2417.13 Ecología Vegetal
2506.16 Teledetección (Geología)
3106.01 Conservación
3106.06 Protección
3106.99 Otras (Incendios forestales)
Descripción
Sumario:[EN] Background Detailed knowledge of the spatial distribution of vegetation fuels is essential for assessing wildfire hazard and behavior, as well as for planning effective management. In southern Europe, the Prometheus project has proposed the differentiation of seven fuel types, but their characterization using remote sensing techniques remains challenging. Here, we propose a two‑phase, innovative methodology for high‑resolution mapping of Prometheus fuel types, integrating complementary remote sensing data and physically based techniques. In the first phase, we estimated the fire‑propagating element (grass, shrubs, and trees) through multispectral imagery and an advanced spectral unmixing technique (multiple endmember spectral mixture analysis—MESMA) to mimic the Prometheus classification system in the field. In the second phase, synthetic aperture radar data, together with a novel LiDAR workflow related to the distribution of leaf area density by fuel vertical strata, were used to classify the corresponding Prometheus fuel type (FT) within each fire‑propagating element (grassland, shrubland, and woodland) by using a random forest classification algorithm. Results Field validation conducted across four sites in the Iberian Peninsula with markedly different environmental conditions and vegetation types showed high performance in the classification of the fire‑propagating element through MESMA (overall accuracy (OA) = 94.58%). The producer’s (PA) and user’s (UA) accuracy for each class (> 90.00%) was consistent with the OA. During the second phase, fuel types in shrublands (FT2 to FT4) and woodlands (FT5 to FT7), together with the fuel type in grasslands (FT1) retrieved directly from MESMA, were classified with high overall performance (OA = 90.27%) as depicted by the validation of the final Prometheus fuel type map from a set of independent field plots. The PA and UA for most individual FTs exceeded 80%. Conclusions The results of this manuscript provide an accurate characterization of the spatial variability of fuel types within the Prometheus classification system across heterogeneous landscapes. The generalizability of the remote sensing methodology proposed, grounded in physical and ecological principles, represents a significant advance for fuel planning in southern European countries