by Luis Eduardo Muñoz Guerrero
ABSTRACT
Maritime traffic near major ports is shaped by a combination of vessel characteristics and environmental conditions, yet the extent to which weather variables systematically influence navigation behavior remains underexplored at the level of individual, publicly available AIS datasets. This study examines the relationship between meteorological conditions (wind, and a wind-derived sea-state proxy) and vessel navigation behavior (speed, course variability, stopping frequency) in the waters surrounding the Port of Piraeus, Greece, and investigates whether this relationship is moderated by vessel type. Using a stratified subsample of the Piraeus AIS Dataset (133 vessels across five type categories; 294,100 one-minute-resampled position reports spanning August, October and December 2017, chosen to capture summer meltemi, autumn-transition and winter-storm wind regimes) spatially and temporally joined to NOAA wind observations, we apply descriptive statistics, correlation analysis, and multiple regression to test whether wind conditions are associated with significant changes in vessel speed, heading variability, and dwell (stopping) behavior, and whether these effects differ across vessel categories (cargo, tanker, passenger, fishing, other). Wind speed showed a negligible zero-order correlation with vessel speed (Pearson r = 0.008) but a small, statistically significant effect once vessel type and its interaction with wind were modeled (R² = 0.072, p < .001): cargo vessels slowed modestly as wind increased, while passenger, tanker and fishing vessels showed a flat-to-slightly-positive wind–speed relationship. Heading variability was weakly and inconsistently associated with wind (R² = 0.013), with the direction of the effect differing by vessel type. The probability of a sustained stop was significantly, if modestly, related to wind speed (pseudo-R² = 0.104, p < .001) and strongly differentiated by vessel type, with harbor/service (“Other”) vessels far more likely to be recorded as stopped than cargo, tanker or fishing vessels. Overall, wind conditions in this sample were consistently significant predictors of navigation behavior at this scale of data, but explained only a small share of total variance, and vessel type was the dominant moderator — supporting a nuanced, type-specific rather than uniform weather-response model of port-approach navigation. This work contributes an applied, reproducible methodology for weather-behavior analysis using open AIS data, with implications for port traffic management and maritime safety planning.
