PM10 Data Assimilation on Real-time Agent-based Simulation using Machine Learning Models: case of Dakar Urban Air Pollution Study

High air pollution is a major health risk. Heavy urbanisation favours the degradation of air quality in large cities such as Dakar. In this city, the annual rate of PM10 exposure in the city is above the threshold recommended by the World Health Organisation (WHO). However, in order to set up a national pollution monitoring network, our approach consists in combining system observations (data from different stations) with a multi-agent simulation. In this paper, we present a model for assimilating PM10 pollution data coupled with a multi-agent realtime simulation. This assimilation model is based on a machine learning method. We performed several simulations to show that the autoregressive ARIMA model is better suited for predicting PM10 pollution data. Then we discussed the relevance of studying other parameters of the model.