Deep Learning Method for Haze Prediction in Singapore

Abstract

In recent years, environmental scientist focused more efforts on studying atmospheric air quality and its relation to global warming. The rapid advancement of deep learning methodology has made it a popular topic for environmental research. With this consideration, we propose a deep learning Recurrent Neural Network (RNN) method to predict the hourly fluctuation of air pollutant associated with the haze phenomena. For this study, we are comparing multi-layer models of stacked RNN and bidirectional RNN. All algorithms tested in this paper were based on either the Long Short-Term Memory Neural Network (LSTM) or Gated Recurrent Unit (GRU). These algorithms are an improvement and enhancement of the existing prediction method done on the basic RNN network. We aim to investigate the effect of stacking additional layers onto prediction model with either LTSM or GRU gates in the hidden network. To compare the overall performance of each method, the mean absolute error (MAE), training and validation loss per epoch are applied to the experiments in this paper. The experimental results indicate that our method is capable of dealing with PM2.5 concentration prediction with the highest performance.