The BEYOND Fire Monitoring Service s an operational platform of continuous operation for detecting active fires across Greece in near real time.

It is based on fully automated processing chains that support early detection and dynamic monitoring of events. It integrates data fusion techniques, modeling, and advanced image processing algorithms developed and optimized through long-term operational use.

The service utilizes high temporal resolution satellite observations from missions such as MSG SEVIRI (EUMETSAT). It also combines atmospheric parameters such as wind, temperature and precipitation (derived from numerical weather predictions), topography, elevation, along with fuel data and risk indices. It produces active fire predictions at a resolution of 350 meters, significantly higher than the original 3.5 km resolution of MSG SEVIRI. The platform further integrates heterogeneous geospatial datasets such as Google Earth, Corine Land Cover, and OpenStreetMap, as well as points of interest, offering a unified operational analysis environment while supporting retrospective fire investigation.

Since summer 2026, the service has been expanded with an ignition point estimation algorithm and fire spread prediction capability per event, utilizing the ForeFire model for dynamic simulation.

The Daily Fire Risk Forecast Map provides, every day around 15:30, an interactive map depicting fire risk for the following day across Greece, with a spatial resolution of 500 meters.

Risk is expressed across five levels, from very low to very high, allowing immediate interpretation. The platform also supports displaying the user’s location on the map via location services.

The forecast is based on advanced artificial intelligence methods and a multi-model approach, using algorithms such as Random Forest, Extra Trees, XGBoost, and neural networks, selected based on performance.

Each estimate relies exclusively on data available up to the previous day and incorporates geospatial and meteorological features such as temperature, wind, and precipitation forecasts, Copernicus DEM, Corine Land Cover, vegetation indices (NDVI, EVI), NASA MODIS land surface temperatures, and historical fire data from the long-term burned area mapping service.

The system’s effectiveness has been validated through extensive testing on past datasets, demonstrating high accuracy in fine-scale risk estimation.

The Diachronic Burned Scar Mapping service was initially developed as an offline multi-sensor processing chain and now operates as an online platform with full coverage of the Greek territory.

It applies multi-temporal analysis to estimate burned areas and environmental impacts from 1984 to the present. It also provides ignition frequency per region, enabling spatial and temporal fire recurrence analysis.

The algorithm is based on an adaptation of BSM-NOA and the methodology of Liu et al. (2014) using Level Set Methods, with additional optimizations for error reduction and heterogeneous satellite data adaptation.

The service uses data from Landsat, SPOT, IKONOS, FORMOSAT, as well as Sentinel-2 and Sentinel-3, ensuring full spatial and temporal coverage, supported by BEYOND infrastructure and the Greek Sentinel Mirror Site.

Biomass burning and forest fires are directly linked to climate change impacts, as increasing global temperatures are intensifying their frequency and severity. Emitted particles play a critical role in climate regulation, affecting radiation transfer and cloud formation processes.

At the same time, exposure to smoke is associated with significant public health impacts, as even highly localized events can affect large regions through particulate transport.

Smoke plume evolution and dispersion are governed by atmospheric factors such as wind fields, atmospheric stability, mixing height, as well as fire intensity, duration, and initial plume rise. Forecasting is implemented through numerical simulations.

Detailed fire hotspot information is provided by the BEYOND 24/7 Fire Monitoring Service, updated every five minutes via MSG SEVIRI, serving as key input for the forecasting models.

Smoke dispersion is then simulated using the FLEXPART Lagrangian dispersion model, driven by meteorological data from the operational WRF-ARW model over Greece.

The Forest Fire Information System of the BEYOND EO Centre is a fully automated, continuously operating system for detecting active fire hotspots and mapping burned areas. Its operational coverage extends beyond Greece to Europe, North Africa, the Black Sea, and the Middle East.

The Centre processes satellite data acquired from the ground receiving station at BEYOND facilities in Penteli, as well as from the Greek Sentinel Mirror Site. Near real-time data processing from satellites such as SUOMI-NPP, NOAA-20, and MODIS provides reliable and up-to-date fire information, supporting fire services and crisis management centers across the EU and neighboring regions.

The system provides active fire maps, hotspot location and extent data, as well as estimates of fire intensity and risk magnitude. It also offers detailed statistics on affected areas based on land use/land cover categories, per fire event, country, and time period.

Long-term data are organized into structured data cubes, enabling further analysis and supporting studies of fire behavior and occurrence probability across the Mediterranean ecosystem.