Data-driven modelling and validation of aircraft inbound-stream at some major European airports

TL;DR

This study compares Poisson and PSRA models for aircraft arrivals at European airports, highlighting the limitations of Poisson due to capacity constraints and demonstrating the effectiveness of a novel non-parametric PSRA approach.

Contribution

The paper introduces a data-driven PSRA model that better captures the correlation structure of aircraft arrivals constrained by finite capacity.

Findings

Poisson model has limitations due to capacity constraints.

PSRA model predicts arrivals more accurately.

Interarrival times may follow exponential distribution.

Abstract

This paper presents an exhaustive study on the arrivals process at eight important European airports. Using inbound traffic data, we define, compare, and contrast a data-driven Poisson and PSRA point process. Although, there is sufficient evidence that the interarrivals might follow an exponential distribution, this finding does not directly translate to evidence that the arrivals stream is Poisson. The main reason is that finite-capacity constraints impose a correlation structure to the arrivals stream, which a Poisson model cannot capture. We show the weaknesses and somehow the difficulties of using a Poisson process to model with good approximation the arrivals stream. On the other hand, our innovative non-parametric, data-driven PSRA model, predicts quite well and captures important properties of the typical arrivals stream.