Modeling Phase-Specific Risk Dynamics in Aviation: A Dual-Axis Analysis of Incident Frequency and Severity

Aviation safety has traditionally focused on take-off and landing—flight phases historically associated with catastrophic accidents. However, the increasing integration of automation in modern aircraft has reshaped risk patterns, with cruise-phase incidents rising in frequency. This study analyzes incident reports from the NASA Aviation Safety Reporting System, employing descriptive statistics, categorical mapping, and temporal trend analysis to quantify phase-specific risk distributions and causation patterns. Cruise phase accounts for the highest proportion of incidents (42.5%), with 95% rooted in technical failures such as autopilot disengagements, sensor anomalies, and avionics malfunctions. Conversely, landing-phase incidents, though rare (5.1%), exhibit heightened risk tied to human decision-making errors (25.5%)—including crosswind misjudgments, delayed go-arounds—and adverse weather interactions. To address these risks, a dual-axis risk mitigation framework (quantifying risk exposure across flight phases) is proposed comprising predictive maintenance by IoT-driven health monitoring of avionics systems to pre-empt cruise-phase failures, human-centered training and scenario-based crew resource management modules targeting landing phase. Temporal analysis further reveals a 14% annual increase in cruise-phase incidents, correlating with rising fleet-wide automation adoption, while landing-phase human errors remain persistently elevated. These findings challenge existing Federal Aviation Administration’s phase-agnostic safety protocols and advocate for the development of phase-specific risk mitigation strategies. The study’s insights have immediate industry applications, including recommendations for next-generation cockpit systems with real-time weather integration and enhanced sensor redundancy standards. Additionally, regulatory reforms, urging the authorities to adopt phase-specific safety protocols tailored to the distinct risk dynamics of cruise and landing, are proposed. Future work will explore machine learning-driven predictive risk analytics and incorporating global datasets for wider generalizability.