Visibility Prediction through Machine Learning: Exploring the Role of Meteorological Factors

The distance at which an observer can identify a black item against a horizontal skyline is described as visibility. The loss of atmospheric visibility in metropolitan areas is caused mainly by solid and liquid particles and gases interfering with the movement of light through the atmosphere. Atmospheric visibility is an essential environmental metric that is affected by weather and air quality. Low visibility situations caused by weather factors create extremely dangerous scenarios on highways, triggering traffic accidents and endangering transportation system operation, performance, and safety. This study aims to investigate the effect of various meteorological parameters to identify the key causes contributing to poor visibility. Additionally, machine learning algorithms will be utilized to model atmospheric visibility. In this study, six international airports in the United Arab Emirates (UAE) were studied for their visibility using several meteorological parameters (air temperature, relative humidity, wind speed, rainfall, sea level pressure, and dewpoint temperature). Furthermore, six Machine Learning algorithms (26 presets) were employed to determine the best model for predicting visibility. The study's finding shows that the Rational Quadratic GPR model is the best model for predicting visibility using several meteorological variables. The developed visibility models are expected to have a wide range of practical applications including but not limited to reducing the likelihood of accidents and improving the efficiency of transportation systems.