Aerosol prediction based on bi-directional LSTM using real observed weather data

Abstract

With increasing of interests in aerosol for national environmental crisis, prevention of aerosol becomes a major issue for human-being health. Several epidemiological studies have shown a clear statistical relationship between aerosol and human mortality [1]. Despite of ongoing efforts, traditional chemical weather forecasting still have many uncertainties. Recently, it was reported that a bidirectional long short-term model (B-LSTM) was able to rebuild non-linear time series data and make valid prediction [2]. Therefore, we propose a B-LSTMbased aerosol prediction model which is suitable for the air quality forecasting.