COMPREHENSIVE STATISTICAL ANALYSIS AND FORECASTING OF SEASONAL FIRE DYNAMICS CONSIDERING CLIMATIC FACTORS

Problem statement. The increase in the number of fires in the Dnipropetrovsk region under climate change conditions poses a serious environmental and socio-economic threat. There is a need for an in-depth analysis of their seasonal dynamics, taking into account meteorological factors, to enable effective forecasting and fire prevention. The purpose of the article is to conduct a comprehensive statistical analysis of fires in the Dnipropetrovsk region in recent years, identify statistically significant factor variables, and develop a mathematical model for fire forecasting, taking into account interannual and seasonal activity as well as natural and climatic factors. Methodology. The study employs one-way and two-way analysis of variance to identify interactions between the factors “month” and “year”, as well as forecasting methods based on the ETS (Exponential Smoothing) model, which accounts for seasonality. Correlation and regression analysis methods were used to construct predictive models for the number of fires. Scientific novelty. A systematic analysis of fire activity in the Dnipropetrovsk region was carried out, taking into account both interannual and intra-annual (seasonal) variability, based on data from 2020 to 2024 and partially from 2025. Using a two-way ANOVA without replication, the nature of variations in the number of fires – undetected by the one-way approach – was clarified. A forecast for the number of fires in 2025 was developed using the ETS model. A division of the annual cycle into three climatically homogeneous groups of months was proposed for correlation-regression analysis, aiming to reveal the specific impact of weather factors in each season. Numerical relationships were established between the number of fires and specific meteorological parameters (temperature, humidity, precipitation, wind speed) for each group of months. Practical significance. The obtained results can be used by local self-government bodies, the State Emergency Service, and environmental agencies to enhance the effectiveness of fire prevention and response in the Dnipropetrovsk region. The developed forecasting models make it possible to estimate the probable number of fires in advance for different periods of the year and under various weather conditions, which facilitates the optimization of resource allocation, planning of preventive measures, public information, and civil protection. Conclusions. Fire activity in the Dnipropetrovsk region demonstrates a clearly defined seasonal pattern, with peaks occurring during the spring and summer months. Two-way ANOVA revealed a statistically significant interaction between year and month, indicating changes in seasonal fire patterns over time. The constructed ETS model provided satisfactory accuracy in forecasting the number of fires for 2025. It was established that meteorological factors affect fire frequency differently across seasons, as confirmed by correlation-regression analysis.