Impact of clustered meteorological parameters on air pollutants concentrations in the region of Annaba, Algeria

The main objective of this study is the characterization of meteorological conditions in the region of Annaba (Algeria) using clustering tools. The proposed two stages clustering approach is based on using the Self-Organizing Maps (SOMs) and the well known K-means clustering algorithm. Quantitative (using two categories of validity indices) and qualitative criteria were introduced to compare and verify the correctness of the results. The different experiments developed, extracted five classes, which were related to typical meteorological conditions in the area. The obtained meteorological clusters are then used to better elucidate the dependency of meteorology on air quality in the presence of seven measured pollutants. In the current paper, Artificial Neural Networks (ANNs), and more precisely, Multi-Layered Perceptron (MLP) is used for modeling air pollutants, as well as, simulating their behaviour in relation to the meteorological parameters of interest. This behaviour is also investigated with the aid of correlation coefficient, where only results are shown for comparison, several relations and conclusions have been drawn.

► The effect of meteorology on air quality is investigated using clustering methods. ► Five clusters have been identified using a two level clustering approach. ► For each cluster an ANN model is designed for modelling air pollution relationships.