Datasets in Space European Territory [ET] Demonstration Data Space

The product include the fuel type and parameters following the standard classification of fuels for the European territory. The final output – fuel classes and static fuel map with the spatial resolution of 1 km2.
Advanced classification models were applied to LiDAR (GEDI and ICESAT-2), multi and hyperspectral satellite imagery (Sentinel-2, Sentinel-3, PRISMA) for fuel type mapping, using standardised datasets from systematic field campaigns, lab experiments, forest inventory data and biophysical models. Surface fuel and canopy metrics (canopy cover, canopy height, canopy base height and canopy bulk density) are parameterised for running propagation potential (Flammap or FCCS). Custom fuel models were described for special interest ecosystems.
Detailed description of algorithm of calculation and data sources available https://edatos.consorciomadrono.es/dataset.xhtml?persistentId=doi:10.21950/YABYCN

Contact: Elena Aragoneses de la Rubia, UAH: e.aragoneses@uah.es

Aragoneses, E., García, M., Salis, M., Ribeiro, L.M., & Chuvieco, E. (2023). Classification and mapping of European fuels using a hierarchical-multipurpose fuel classification system. Earth System Science Data, 15, 1287–1315. https://doi.org/10.5194/essd-15-1287-2023, 2023

The European fuel map is a raster layer representing the first-level fuel types of the FirEUrisk fuel classification system for the continental scale. The European fuel map was generated through the integration of existing land cover datasets and bioclimatic modelling. Then, it was smoothed and resampled to the target spatial resolution (1 km). Finally, it was validated using LUCAS (Land Use and Coverage Area frame Survey), Google data, and the GlobeLand30 map. Further information can be found in the Product User Manual (PUM).

Contact details of the developer: Elena Aragoneses (e.aragoneses@uah.es), University of Alcalá (UAH)

Aragoneses, E., García, M., Salis, M., Ribeiro, L. M., and Chuvieco, E.: Classification and mapping of European fuels using a hierarchical, multipurpose fuel classification system, Earth System Science Data, 15, 1287–1315, https://doi.org/10.5194/essd-15-1287-2023, 2023.

This is a Radiative Transfer Model (RTM)-based Live Fuel Moisture Content (LFMC) product based on the Sentinel-3 Synergy and Sentinel-2 L2A surface reflectance product and a Look-Up Table (LUT)-based inversion procedure.

This product is based on the inversion of Look-Up Tables (LUT) created using a combination of the PROSPECT-D, 4SAIL RTMs and a modified version of Huemmrich’s GeoSail, and convolving the calculated reflectances in the Sentinel-3 SYNERGY and Sentinel-2 MSI channels. The LUTs change depending on the land cover type: PROSPECT-D and 4SAIL are used for grasslands and shrublands, while for forests the 4SAIL output is fed into the Geo component to simulate discontinuous vegetation. The LUT inversion is performed using Random Forest Regression algorithms. Latest publication available here: Pampanoni, Valerio, et al. "Early Validation of A Live Fuel Moisture Content Product Based on Sentinel-2 and Sentinel-3 Images." IGARSS 2023-2023 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2023. https://doi.org/10.1109/IGARSS52108.2023.10281970.

Pampanoni, Valerio, et al. "Using Prosail Look-Up Tables to Train Random Forests Regressors for Fast Live Fuel Moisture Retrieval." IGARSS 2024-2024 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2024. https://doi.org/10.1109/IGARSS53475.2024.10642386.

Yebra, M., Dennison, P., Chuvieco, E., Riaño, D., Zylstra, P., Hunt, E.R., Danson, F.M., Qi, Y., & Jurdao, S. (2013). A global review of remote sensing of live fuel moisture content for fire danger assessment: moving towards operational products Remote Sensing of Environment 136, 455-468.

The product estimates the probability that a human caused fire will be ignited. The product was developed using Random Forest algorithm based on socio-economic and climatic drivers of ignition from geospatial indicators and demographic databases. The model obtained a performance of 80% estimated by the area under the curve (AUC). The model uses all ignitions that generated burned areas > 100 ha for a total of 33,388 ignitions. In addition, different data balancing methods were tested for the training, the stratified method being the most suitable for this model. This method consists of using a descriptive layer of European ecological regions to obtain a more representative sample of absence points. The results of the climate model showed that the probability of ignition in northern Europe is masked, the human models encountered challenges with classification accuracy, suggesting that the RF algorithm struggled to make accurate predictions based solely on human variables. These findings imply that predicting large fires initiations across broad geographic extents may not be feasible with human variables alone, highlighting the necessity of including climatic variables. This limitation depends on the scale of the analysis, as it dictates the magnitude of potential environmental gradients that drive initiation patterns. Whereas the mixed model proved to be the best candidate to represent the probability at the European scale. This might be because climatic variables help the model to differentiate between northern and southern Europe and thus find patterns in the data. It is important to note that by using the mixed models, areas of high probability are seen in northern Europe, which would otherwise be masked. Because of the results, we suggest the use of mixed models for global studies. For further information about the methodology and variables used see (Ochoa et al 2024)

Ochoa, C., Bar-Massada, A., & Chuvieco, E. (2024). A European-scale analysis reveals the complex roles of anthropogenic and climatic factors in driving the initiation of large wildfires. Science of the total environment, 170443. https://doi.org/10.1016/j.scitotenv.2024.170443

The Fire Forecasting Model based Natural Ignition Probability product predicting the wildfires ignition probability by natural causes through basic meteorological parameters provided by Numerical Weather Predicting models. The product was developed within the Activity “A1.1.2 Natural fire ignitions: Lightning”.
FFM employs a multi-step machine learning procedure to construct a statistical model that predicts Fire Radiative Power (FRP) based on global ERA5 reanalysis

(https://www.ecmwf.int/en/forecasts/dataset/ecmwf-reanalysis-v5),
ERA5-Land reanalysis (https://www.ecmwf.int/en/era5-land),
IFS operational forecast (https://www.ecmwf.int/en/forecasts/documentation-and-support/changes-ecmwf-model),
Terra's Moderate Resolution Imaging Spectroradiometer (MODIS, MOD14/MYD14, collection 6, https://www.earthdata.nasa.gov/learn/find-data/near-real-time/firms).

The model utilises various meteorological parameters, including Cloud-to-Ground Lightning Flash Density and Fire Danger Indices, as predictors for training and making predictions by calculating their respective contributions to total FRP. The contribution of cloud-to-ground lightning flash density to FRP is then used as a proxy for natural ignition probability.

Contact details of the developer: evgeny.kadantsev@fmi.fi

This cartographic product estimates Exposure based on a simple approach that considers the resources potentially exposed to wildfire effects. Even though different methods have been proposed to estimate this variable (see deliverable D.1.4 and (Chuvieco et al. 2023), within this version of the integrated risk component only three categories of exposure were considered, given them a certain weight in relation to the importance of being potentially affected by fire:

• Unburnable (code 0), including water, bare soil and urban dense.
• Burnable (code 0.8), including all vegetation categories that are susceptible of being burned.
• Burnable and with Wildland Urban Interface (WUI (Code 1). Includes burnable categories within a radius
of 1 km of an urbanised area.

For mapping the WUI at ET scale, first the extension of the urbanised area was mapped, based on the 2019 World Settlement Footprint dataset (WSF). This dataset was generated by automated interpretation of 10-m resolution Sentinel-1 and Sentinel-2 satellite imagery, and each pixel is either built-up or not. Next, vegetation data was gathered from the 2020 European Space Agency (ESA) global land cover product (Zanaga et al. 2020). This dataset is a classification of the same 10-m Sentinel-1 and Sentinel-2 satellite imagery, in conjunction with auxiliary data, into 11 land cover classes. This dataset was reclassified into two classes: woody vegetation classes (‘forests’ and ‘shrublands’, “1”) and others (“0”). Hence, the research focused on fire hazard only from woody vegetation, regardless of species identity and fuel characteristics.

The most common approach for WUI mapping distinguishes between WUI intermix and interface classes (Radeloff et al. 2005). An intermix WUI is a unit area that: 1: comprises enough buildings exposed to wildfire risk; and 2: contains or is directly surrounded by sufficient amounts of flammable vegetation. An interface WUI is a unit area that contains buildings that are not directly surrounded by flammable vegetation yet is within a relatively short distance from a contiguous patch of flammable vegetation which can produce firebrands. To map the WUI, the parameters of the existing point-based WUI mapping approach were modified (Bar-Massada et al. 2013), which is based on the original US WUI mapping approach.

The new approach adapts the 10 m-cell grid of the underlying datasets and includes the following steps:

1. Including all cells within 100 m of a built-up cell as candidate WUI locations, regardless of the housing
density within (and around) them in order to identify all built-up areas that are potentially at risk.
2. Calculating the percent cover of woody vegetation within a 500 m radius around candidate WUI location
cells; if woody cover was greater than 50%, these cells were mapped as ‘intermix WUI’.
3. Identifying patches of woody vegetation that were larger than 5 km2.
4. Identifying all candidate WUI locations that were within 600 m to large vegetation patches; these cells
were mapped as ‘interface WUI’.

The choice of 600 m reflects findings about the approximate median value of maximum travel distances of flying embers. This threshold differs from the previously used distance of 2,400 m (Carlson et al. 2022; Radeloff et al. 2005), which reflects maximum travel distances. The result of this process is a WUI map at either PS scale (if built-up information was based on individual buildings; five WUI maps overall) or ET scale (when the input was the WSF dataset; one WUI map for the entire ET). Based on these maps, we quantified the overall extent and the spatial distribution of interface, intermix, and total WUI at the country level for each country in our study area. We also calculated the variation in WUI cover within countries, by overlaying a map of statistical regions at NUTS-3 level obtained from Eurostat, the EU’s statistical agency.

Contact: Clara Ochoa, UAH: mclara.ochoa@uah.es or Avi Bar-Massada, HU: avi-b@sci.haifa.ac.il


A. R. Carlson, D. P. Helmers, T. J. Hawbaker, M. H. Mockrin, and V. C. Radeloff, “The wildland–urban interface in the United States based on 125 million building locations,” Ecological Applications, vol. 32, no. 5, p. e2597, Jul. 2022, doi: https://doi.org/10.1002/eap.2597.

E. Chuvieco et al., “Towards an integrated approach to wildfire risk assessment: when, where, what and how may the landscapes burn,” Fire, vol. 6, no. 5, p. 215, 2023.

A. Bar-Massada, S. I. Stewart, R. B. Hammer, M. H. Mockrin, and V. C. Radeloff, “Using structure locations as a basis for mapping the wildland urban interface,” J Environ Manage, vol. 128, pp. 540–547, 2013, doi: https://doi.org/10.1016/j.jenvman.2013.06.021.

V. C. Radeloff, R. B. Hammer, S. I. Stewart, J. S. Fried, S. S. Holcomb, and J. F. McKeefry, “THE WILDLAND–URBAN INTERFACE IN THE UNITED STATES,” Ecological Applications, vol. 15, no. 3, pp. 799–805, Jun. 2005, doi: https://doi.org/10.1890/04-1413.

D. Zanaga et al., “ESA WorldCover 10 m 2021 v200.” [Online]. Available: https://zenodo.org/records/7254221

The product is a raster TIFF layer with a spatial resolution of 1 km, providing current ecological value data across Europe. This dataset offers up-to-date information, making it suitable for contemporary environmental analyses and applications.

The methodology used for ecological values assessment (EVA) is based on characterising the ecosystems' biological distinctiveness (BD) and conservation status (CS) based on the methodology scheme proposed by Dinerstein et al., (1995. We stated that ecosystems hosting high taxonomic richness or rare plant communities, or habitats of endangered species highly contribute to their value through their enhanced functioning. The destruction or degradation of these valued ecological components, notably by fires, can have long-term implications for ecosystem health and conservation.

The previously described indicators (from BD and CS) were introduced in the Principal Components Analyses (PCA) model [24], taking the positive direction for values close to 1000 and negative for those close to 0. Species richness, forest productivity, density of habitat, key biodiversity areas, exceptional forest, unique habitat preservation and places of special conservation conserve the original direction. In contrast, the human pressure and loss of forest indicators inverse the direction, taking values close to 1000 where human pressure and loss of forest values were low. Regarding the previous explanation, we selected the first PCA because it represents the synergies between conservation status and biological distinctiveness. For example, zones with fewer roads and railways and a loss of forest cover conserve better ecological values.

Contact:Emilio Chuvieco, UAH: emilio.chuvieco@uah.es

Arrogante-Funes, F., Mouillot, F., Moreira, B., Aguado, I., & Chuvieco, E. (2024). Mapping and assessment of ecological vulnerability to wildfires in Europe. Fire Ecology, 20(1), 98. https://doi.org/10.1186/s42408-024-00321-8

E. Dinerstein et al., “A conservation assessment of the terrestrial ecorregions of Latin America and the Caribbean,” Banco Mundial, Washington DC, Sep. 1995. Accessed: Feb. 13, 2025. [Online]. Available: http://documents.worldbank.org/curated/en/957541468270313045

This product estimates the economic values of expected damages by wildfires for the European Territory assessed for different assets (some provisioning services such as agricultural and forestry goods; regulating ecosystem services such as pollination, carbon sequestration and soil erosion; and manufactured capital such as properties). Values are mapped and reported at a resolution of 1 km2 for the year 2021.

The approach used for the assessment of the potential damage of wildfires is based on the correction of the economic value of natural capital by a loss coefficient (ranging 0 to 1), function of the intensity of fire (Dinerstein et al 1995) and the time of recovery of the asset (Brun et al 2019). Coefficients of loss are measured in Spanish Mediterranean forests for timber (Sritharan et al 2022), carbon sequestration (Anderson et al 2009) and soil erosion (Arrogante-Funes et al 2024) and adapted to other assets (agricultural perennial crops) according to the judgement of the research team.
Economic values of natural capital (agricultural and forestry assets) are assessed by market-based approach. The values of regulating ecosystem services are taken by the INCA project (Archibald et al 2019, Vargas et al 2023), while the value of properties is made by correcting national prices provided by Numbeo by the local GDP per capita (Shwörer et al 2014).

We derived maps of damage for timber, livestock, fruit trees, olive groves and vineyards, assuming that the value of natural capital asset is lost for the time necessary for a specific asset to recover from wildfires. Recovery time for trees is variable from 50 to 200 years (Vargas et al 2023). We have assumed 50 years as a reasonable time to assess the damage of timber. There is no information in the literature about the recovery time of fruit trees, olive groves and vineyards, therefore the choice of recovery time (5 years) was guided only by judgement of the research team. Recovery time for the damage of ecosystem services such as soil retention are measured assuming values proposed by Milazzo et al. (2022) (Stevens et al 2020), ranging from 6 to 20 years.

These damages are assessed as the present value of the future losses, assumed constant until the natural asset has fully recovered (Brun et al 2019). A social discount rate of 3.5% is used.

Values of carbon sequestration are generated with an approach that considers the recovery of vegetation following a logistic curve (Castillo et al 2017). The parameters of the logistic curve are not empirically estimated, nor deduced from the literature, but proposed by the research team to test their sensitivity on the expected damage.
Damage to properties is assessed by correcting the property values by a coefficient of loss, which is function of the material of the property (concrete or wood) and the intensity of fire. These coefficients of loss are proposed by the research team but not yet validated.

Contact: Simone Martino, HUTTON: simone.martino@hutton.ac.uk
Clara Ochoa, UAH: mclara.ochoa@uah.es


T. Andersen, J. Carstensen, E. Hernández-García, and C. M. Duarte, “Ecological thresholds and regime shifts: approaches to identification,” Trends Ecol Evol, vol. 24, no. 1, pp. 49–57, 2009, https://doi.org/10.1016/j.tree.2008.07.014.

S. Archibald, G. P. Hempson, and C. Lehmann, “A unified framework for plant life-history strategies shaped by fire and herbivory,” New Phytologist, vol. 224, no. 4, pp. 1490–1503, Dec. 2019, https://doi.org/10.1111/nph.15986

F. Arrogante-Funes, F. Mouillot, B. Moreira, I. Aguado, and E. Chuvieco, “Mapping and assessment of ecological vulnerability to wildfires in Europe,” Fire Ecology, vol. 20, no. 1, p. 98, 2024, https://doi.org/10.1186/s42408-024-00321-8

M. E. Castillo, J. R. Molina, F. Rodríguez y Silva, P. García-Chevesich, and R. Garfias, “A system to evaluate fire impacts from simulated fire behavior in Mediterranean areas of Central Chile,” Science of The Total Environment, vol. 579, pp. 1410–1418, 2017, https://doi.org/10.1016/j.scitotenv.2016.11.139.

E. Dinerstein et al., “A conservation assessment of the terrestrial ecorregions of Latin America and the Caribbean,” Banco Mundial, Washington DC, Sep. 1995. Accessed: Feb. 13, 2025. [Online]. Available: http://documents.worldbank.org/curated/en/957541468270313045

C. Schwörer, P. D. Henne, and W. Tinner, “A model-data comparison of Holocene timberline changes in the Swiss Alps reveals past and future drivers of mountain forest dynamics,” Glob Chang Biol, vol. 20, no. 5, pp. 1512–1526, May 2014, https://doi.org/10.1111/gcb.12456.

M. S. Sritharan, B. C. Scheele, W. Blanchard, C. N. Foster, P. A. Werner, and D. B. Lindenmayer, “Plant rarity in fire-prone dry sclerophyll communities,” Sci Rep, vol. 12, no. 1, p. 12055, 2022, https://doi.org/10.1038/s41598-022-15927-8
J. T. Stevens, M. M. Kling, D. W. Schwilk, J. M. Varner, and J. M. Kane, “Biogeography of fire regimes in western US conifer forests: A trait‐based approach,” Global Ecology and Biogeography, vol. 29, no. 5, pp. 944–955, 2020.

P. Vargas, R. Heleno, and J. M. Costa, “EuDiS - A comprehensive database of the seed dispersal syndromes of the European flora,” Biodivers Data J, vol. 11, p. e104079, Jul. 2023, [Online]. Available: https://doi.org/10.3897/BDJ.11.e104079

This cartographic product represents estimated property values across Europe in €/ha (2021 currency), integrating national property prices with local economic data. Given the lack of detailed property value databases, the estimation is based on a national-scale model, assuming a linear relationship between property value and regional income. The map is built using Numbeo’s national property prices (outside city centers), adjusted with EUROSTAT GDP per capita ratios (2019) at national and NUTS2 levels. To refine spatial distribution, a 2015 gridded GDP dataset was applied. Additionally, potential fire damage to properties is assessed using loss coefficients based on construction material (concrete or wood) and fire intensity, though these coefficients remain unvalidated. The product provides a spatialised representation of property values and potential economic losses due to fire across Europe.

The value of properties is made by correcting national prices provided by Numbeo by the local GDP per capita. Details are provided by Martino et al. (2023). The property values are expressed in €/ha (2021 currency). Due to limited property data, values were estimated at a national scale using DGP maps, assuming a linear relation between property value and regional income. The baseline was the national average property value outside city centres from Numbeo website, adjusted using EUROSTAT GDP per capita ratios (2019) at both national and NUTS2 levels. To map the values of properties in Europe, a 2015 GDP dataset was used, adjusting local GDP by national GDP and applying spatial corrections. Numbeo values, originally in dollars, were converted to euros using 2021 exchange rates (0.84). Damage to properties is assessed by correcting the property values by a coefficient of loss, which is function of the material of the property (concrete or wood) and the intensity of fire. These coefficients of loss are proposed by the research team but not yet validated.

Contact: Simone Martino, HUTTON: simone.martino@hutton.ac.uk
Clara Ochoa, UAH: mclara.ochoa@uah.es

The product integrates the different components of risk developed within FirEUrisk to give a synthetic view of fire risk conditions for a specific moment and area. The input data includes static data (human ignition, exposure…) and dynamic components (weather conditions, fuel moisture…). The table below includes a list of inputs. The overview figure below shows a graphical representation of the different components of risk considered in FirEUrisk. The Inputs to compute the Integrated Risk Index, listed below.
Variable Units Sources
TP, Temperature º C DWD-ICON7 data
Pp, Precipitation mm DWD-ICON7 data
WS, Wind speed m/s DWD-ICON7 data
RH, Relative Humidity % DWD-ICON7 data
FT, Fuel types Fuel classes FirEUrisk fuel map
PNI, Probability of natural ignition 0/1000 Weather data and lightning
PHI, Probability of human ignition 0/1000 Human and biophysical factors
LFMC, Live fuel moisture content % DW Satellite data
E, Exposure 0, 0.8, 1 WUI, land cover
SEV, Value of Houses and Infrastructures €/ha Statistics
CL, Climatic regions 1-2 /No-Yes Med Proxy of building materials
EST, Ecosystem service values. €/ha/year Statistics
EVa, Ecological values 0/1000 Cartographic & vegetation databases
LEV, Potential loss of Ecological values 0/1000 Coping capacity, resistance
RT, Regeneration time years Several species
The integration of the different components was based on quantitative weighting, considering the relative importance of each component in terms of extreme fire risk assessment. The integrated risk index (IRI), was computed by combining the Danger (D), Exposure (E) and Vulnerability (V), with the following formula:
IRI = (0.6*D + 0.4*V)*E
Danger was derived from Probability of Ignition and Propagation Potential. Vulnerability was derived from estimated social damages (including housing and ecosystem services) and ecological damages (including the coping capacity of plants and their recovery potential), by estimating the reduction of values (both ecosystem services and ecological values) that would occur in case the area is burned. The conditions of that burning depend on propagation estimates, considering weather conditions and fuel characteristics, particularly on Fire Line Intensity and Flame Length, which affect the degree of estimated post-fire severity. Exposure was estimated by considering burnable covers (weight 0.8), with an additional weight for those areas with Wildland-urban interface (weight 1), having 0 the unburnable areas.

Contact: Emilio Chuvieco, UAH: emilio.chuvieco@uah.es

Chuvieco, E., Yebra, M., Martino, S., Thonicke, K., Gómez-Giménez, M., San-Miguel, J., Oom, D., Ramona Velea, Florent Mouillot, Juan R. Molina, Ana I. Miranda, Diogo Lopes, Michele Salis, Marin Bugaric, Mikhail Sofiev, Evgeny Kadantsev, Ioannis Gitas, Dimitris Stavrakoudis, George Eftychidis, Bar-Massada, A., Alex Neidermeier, Valerio Pampanoni, Pettinari, M.L., Arrogante, F., Ochoa, C., Moreira, B., & Viegas, D. (2023). Towards an integrated approach to wildfire risk assessment: when, where, what and how may the landscapes burn. Fire, 6, 215, https://doi.org/10.3390/fire6050215