Forecasting Smog Clouds With Deep Learning

TL;DR

This study explores deep learning models, especially hierarchical GRU architectures, for multivariate air pollution forecasting using meteorological data from two locations, demonstrating their effectiveness and efficiency.

Contribution

Introduces a hierarchical, multi-task deep learning model inspired by atmospheric science for improved smog cloud forecasting.

Findings

Hierarchical GRU outperforms other models in accuracy.

Multi-task learning enhances forecast robustness.

Model is computationally efficient.

Abstract

In this proof-of-concept study, we conduct multivariate timeseries forecasting for the concentrations of nitrogen dioxide (NO2), ozone (O3), and (fine) particulate matter (PM10 & PM2.5) with meteorological covariates between two locations using various deep learning models, with a focus on long short-term memory (LSTM) and gated recurrent unit (GRU) architectures. In particular, we propose an integrated, hierarchical model architecture inspired by air pollution dynamics and atmospheric science that employs multi-task learning and is benchmarked by unidirectional and fully-connected models. Results demonstrate that, above all, the hierarchical GRU proves itself as a competitive and efficient method for forecasting the concentration of smog-related pollutants.