Forecasting of Satellite Based Carbon-Monoxide Time-Series Data Using a Deep Learning Approach
Abhishek Verma, Virender Ranga, Dinesh Kumar Vishwakarma
Abstract
In last few decades one of the major problems is air pollution which has raised the eyebrows of everyone. Despite all the efforts, it still lies in the category of dangerous. In air pollution there is one of the most hazardous gases named carbon monoxide which is a matter of concern & produced mostly whenever a material burns with a lack of oxygen. This paper presents the forecasting of carbon monoxide with the help of a satellite-based sentinel 5p dataset using earth engine. Further, with the help of the deep learning approach ‘LSTM’, we forecast a time series base result. We have trained and tested the data using a deep-learning model. We have evaluated the potential results by overlapping the original and predicated values and calculating Root-mean-square (RMS) error to validate our approach. The results show that the method of LSTM is very efficient and accurate.