Natural-Language-Processing-Enabled Quantitative Risk Analysis of Aerial Wildfire Operations

Abstract

Aerial wildfire operations are high risk and account for a large number of firefighter deaths. The increasing intensity of wildfires is driving a surge in aerial operations, as well as interest to improve system safety and performance. In this work, wildfire aviation mishaps documented using the Aviation Safety Communiqué (SAFECOM) system are analyzed using a previously developed framework for hazard extraction and analysis of trends. Hazards and specific failure modes are extracted from the narrative data in SAFECOM forms using natural language processing techniques. Metrics for each hazard (including the frequency, rate, and severity) are calculated. We examine whether these metrics change over time and whether they are related to metadata, such as region and aircraft type. The results of the hazard analysis are presented in a risk matrix, identifying the highest and lowest risk hazards based on the rate of occurrence and average severity. The analysis of all SAFECOM reports indicated that the jumper operations hazards were classified as high risk; whereas the hydraulic fluid malfunctions, bucket or tank failures, retardant loading and jettison failures, prescribed burn operations, cargo letdown failures, and severe weather were classified as serious risk. However, when applied to a specific operational scenario, risk levels change across hazards.