Collision Risk Modeling In The Northern Pacific Airspace Under Separation Reduction And Improvement In Navigational Performance PDF Download

"Conventional wisdom suggests aircraft midair collisions to be random events, governed by the laws of Brownian Motion, and best analyzed by stochastic methods. An alternative hypothesis, that such accidents are deterministic in nature, and that specific factors leading to midair collisions can be identified and mitigated, forms the basis for this Dissertation. A predictive model using case control theory is developed for assessing Risk Index, a criterion measure of midair collision likelihood, for any General Aviation flight, actual or hypothetical. Generating the model requires statistical validation of two independent near midair collision databases, and identifying within them those aircraft, aircrew and airspace characteristics most closely associated with collision risk. Calibration of the model shows reality to fall somewhere between the stochastic and deterministic assumptions. A statistically significant correlation is found between predicted and observed Risk Index for a sizable random sample of flights, with a resulting Coefficient of Determination of 0.25. This suggests that we have identified 25% of the source of variance in midair collision risk, the remaining 75% being random. Therefore we can realistically hope to reduce midair collisions by roughly 25%. Strategies for mitigating the identified causal factors are proposed. Measures to reduce the random, remaining 75% of collision risk are also explored. However, these appear to require a significant overhaul of Air Traffic Control procedures, which must be approached with caution, to guard against the attendant possibility of curtailing capacity in the Air Transportation System."--Page 1-2