Forecasting PM10 Levels Using Machine Learning Models in the Arctic: A Comparative Study

In this study, we present a statistical forecasting framework and assess its efficacy using a range of established machine learning algorithms for predicting Particulate Matter (PM) concentrations in the Arctic, specifically in Pallas (FI), Reykjavik (IS), and Tromso (NO). Our framework leverages historical ground measurements and 24 h predictions from nine models by the Copernicus Atmosphere Monitoring Service (CAMS) to provide PM10 predictions for the following 24 h. Furthermore, we compare the performance of various memory cells based on artificial neural networks (ANN), including recurrent neural networks (RNNs), gated recurrent units (GRUs), long short-term memory networks (LSTMs), echo state networks (ESNs), and windowed multilayer perceptrons (MLPs). Regardless of the type of memory cell chosen, our results consistently show that the proposed framework outperforms the CAMS models in terms of mean squared error (MSE), with average improvements ranging from 25% to 40%. Furthermore, we examine the impact of outliers on the overall performance of the model.

Automated summary

In this study, a statistical forecasting framework is proposed and evaluated using various machine learning algorithms to predict Particulate Matter (PM) concentrations in the Arctic region (Pallas in Finland, Reykjavik in Iceland, and Tromso in Norway). The framework utilizes historical ground measurements and 24-hour predictions from Copernicus Atmosphere Monitoring Service (CAMS) models to provide PM10 predictions for the next 24 hours. Different memory cells based on artificial neural networks are compared, and the proposed framework consistently outperforms the CAMS models in terms of mean squared error (MSE), with average improvements ranging from 25% to 40%. The impact of outliers on the overall performance of the model is also examined.