Predicting particulate matter concentrations based on atmospheric conditions -

Abstract

Research in Air quality Monitoring has been gaining a great importance worldwide especially in areas where pollution levels are high. The main objective of this thesis is to develop a computer model to predict ground pollution levels based on meteorological conditions. In order to build this model, daily mixing height data were used, estimated from temperature profiles collected from a simulator (WRF) for the period of nine months. The analysis was performed over the region of Beirut and involved the usage of pollution parameters, such as the PM2.5, PM4, and PM10 concentrations which were measured for this period, and meteorological parameters such as the mixing height, relative humidity, and wind speed. The study confirmed that there is strong anti-correlation between the mixing height and near ground level PM concentrations (PM 2.5 and PM10), moderate positive correlation between the relative humidity and near ground level PM concentrations, and weak negative correlation between the wind speed and near ground level PM concentrations. Regression models produced good results for the mixing height as a predictor for PM2.5 and PM10 concentrations. The mixing height was the most dominant factor in the regression analysis among other meteorological parameters including Relative Humidity and Wind Speed. Multi variable regression models (depending in two and three independent variables) were developed to predict PM concentrations based on meteorological parameters. The best regression coefficients were witnessed with the multi variable regression models developed to predict PM concentrations based on the three meteorological parameters (mixing height, relative humidity, and wind speed). These models can be applied for prediction of near ground pollution level over for the region of Beirut.