Voronezh State University of Forestry and Technologies named after G.F. Morozov has built a system that combines satellite data, drone imagery, and machine learning to predict fire-prone forest areas, forecast fire spread, and calculate carbon losses. The system includes a model that assesses the condition of tree stands and estimates combustible materials like fallen needles and deadwood, then automatically assigns a natural fire danger class to each forest stand. It also introduces a method to calculate carbon stored in forest ecosystems and combustion coefficients for different fire types, enabling pre-fire risk assessment and post-fire carbon loss estimates. Developers note the fire season in European Russia has lengthened by two to four weeks in recent years, with Russia seeing 9,000 to 35,000 forest fires annually covering up to several million hectares. Average yearly damage is about 20 billion rubles, according to Rosleskhoz. The tools are designed to plug into existing forest fire monitoring systems, with the next step being remote AI-based assessment of forest conditions so agencies can act before a fire starts. The system's carbon loss methodology could support climate reporting and insurance products, though validation against real fire events will be critical.
