Fires and fire risk perception predictors on Finnish farms

The physical and operational farm environment has become more complex, challenging risk management. This study aimed to identify the number and typical features of farm fires and the factors connected to farmers' assessments of fire risk. The data comprised a subset extracted from the national accident database of the Finnish rescue services from 2000 to 2021 (3,989 farm fire alarm calls) and farm survey data collected in the spring of 2021 (506 valid responses). The descriptive analysis of the fire incident data showed that the most typical fire sites on farms were dairy cattle buildings, grain dryers, and heating plants. Most farmers (83.4%) assessed the probability of fire as intermediate or high. The Bayesian machine learning approach used to analyse the survey data suggested the control of malfunctions, failures of machinery and equipment, and educational efforts were key elements in fire risk management.