Forecasting Low Horizontal Visibility in Murtala Muhammed Airport, Ikeja, Lagos, Nigeria

Abstract

Air transport is recognized worldwide as the quickest, safest and most reliable means of movement from continent to another. Although aviation is the leading sector in transportation it still has some challenges to overcome. Air transportation calls for a need to understand the atmosphere as it significantly affects its operation; hence the undisputable need of meteorology (a branch of science dealing with the earth's atmosphere and the physical processes occurring in it) in aviation. Weather affects aviation operations and could be improved upon by monitoring and understanding the relationship of weather variables. Accurate weather observation is very important for it is the back-bone of weather forecasting as present weather records serves as input variables upon which forecast models are built, hence accurate reports should be made. Weather forecasting is of paramount importance in aviation operation at Murtala Muhammed International Airport. This research is aim at generating a technique that will aid in predicting low visibility at the airport. The result shows that four major weather conditions account for the occurrence of low visibility. They are haze, rain, fog and mist. Result of correlation analysis reveals that development of these weather conditions are favoured by two major causative factors which are relative humidity (RH) and dew point temperature. Days of high RH tends to bring about low visibility while reduction in dew point temperature favour low visibility occurrence. Hence, it is reasonable to conclude on the effectiveness of using these two causative factors in developing predictive model for low visibility in order to get accurate forecast of low visibility in the study area it is of paramount important to also consider values of dew point temperature and relative humidity as they are the main causative factors of low visibility in the airport. The use of models such as multilayer perceptron which uses artificial intelligence to learn and recognize pattern should be employ in weather forecasting. Multilayer Perceptron (MLP) and multiple linear regression analysis are the two predictive model used and it shows that MLP generated a model which is good for predicting low visibility for haze, fog and mist conditions while Multiple Linear Regression model is good for predicting low visibility during low visibility on rainy conditions. Although from the result it showed that it MLP behaves poorly during rainy conditions compare to regression model, therefore forecasters should not rely on just one model but should employ the use of different models in low visibility forecasting.