Convolutional Neural Network For Short Term Fog Forecasting Based On Meteorological Elements

Fog is the main weather phenomenon that causes low visibility, which makes traffic and outdoor work extremely dangerous. It is urgent to improve the accuracy of fog forecast. In this paper, ground observation meteorological elements time series data is converted into 2D image format, then we train a simple convolution neural network to predict the existing of short time fog. Different experiments is arranged to validate the performance of the proposed method, which obtained the best prediction recall 71.43% and 71.47% for next four and two hours respectively. Contrasting traditional numerical prediction and model prediction method, the application of convolutional neural network method to fog prediction is our first attempt.