Forecasting Smog Clouds With Deep Learning: A Proof-Of-Concept

Oldenburg, Valentijn (2024) Forecasting Smog Clouds With Deep Learning: A Proof-Of-Concept. Bachelor's Thesis, Artificial Intelligence.

Preview	Text bAI2024OldenburgVW.pdf.pdf Download (3MB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (159kB)

Abstract

Air pollution and smog carry correlations to numerous pervasive health effects. Given the risks, foreseeing toxic pollutant levels poses a vital challenge that, upon resolution, enacts a framework for life-saving decisions. Data-driven deep learning (DL) methods offer a novel approach to air quality prediction, yet their potential for modelling a combined set of smog-related pollutants with recurrent neural nets (RNNs) remains unexplored. In this proof-of-concept study, we conduct multivariate timeseries forecasting of nitrogen dioxide (NO2), ozone (O3), and (fine) particulate matter (PM10 & PM2.5) with meteorological covariates between two locations in the Netherlands using various DL architectures, with a focus on RNNs with long short-term memory (LSTM) and gated recurrent unit (GRU) memory cells. In particular, we propose an integrated, hierarchical model architecture inspired by air pollution dynamics and atmospheric science that employs multi-task learning (MTL) 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 smog-related pollutants.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Cardenas Cartagena, J. D.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	23 May 2024 12:34
Last Modified:	23 May 2024 12:34
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/32424

Actions (login required)

View Item