Interpretable Air Pollution Forecasting by Physics-Guided Spatiotemporal Decoupling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Zhiguo, Ma, Xiaoliang, Schlesinger, Daniel
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911391320899584
author Zhang, Zhiguo
Ma, Xiaoliang
Schlesinger, Daniel
author_facet Zhang, Zhiguo
Ma, Xiaoliang
Schlesinger, Daniel
contents Accurate and interpretable air pollution forecasting is crucial for public health, but most models face a trade-off between performance and interpretability. This study proposes a physics-guided, interpretable-by-design spatiotemporal learning framework. The model decomposes the spatiotemporal behavior of air pollutant concentrations into two transparent, additive modules. The first is a physics-guided transport kernel with directed weights conditioned on wind and geography (advection). The second is an explainable attention mechanism that learns local responses and attributes future concentrations to specific historical lags and exogenous drivers. Evaluated on a comprehensive dataset from the Stockholm region, our model consistently outperforms state-of-the-art baselines across multiple forecasting horizons. Our model's integration of high predictive performance and spatiotemporal interpretability provides a more reliable foundation for operational air-quality management in real-world applications.
format Preprint
id arxiv_https___arxiv_org_abs_2511_20257
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Interpretable Air Pollution Forecasting by Physics-Guided Spatiotemporal Decoupling
Zhang, Zhiguo
Ma, Xiaoliang
Schlesinger, Daniel
Machine Learning
Artificial Intelligence
Accurate and interpretable air pollution forecasting is crucial for public health, but most models face a trade-off between performance and interpretability. This study proposes a physics-guided, interpretable-by-design spatiotemporal learning framework. The model decomposes the spatiotemporal behavior of air pollutant concentrations into two transparent, additive modules. The first is a physics-guided transport kernel with directed weights conditioned on wind and geography (advection). The second is an explainable attention mechanism that learns local responses and attributes future concentrations to specific historical lags and exogenous drivers. Evaluated on a comprehensive dataset from the Stockholm region, our model consistently outperforms state-of-the-art baselines across multiple forecasting horizons. Our model's integration of high predictive performance and spatiotemporal interpretability provides a more reliable foundation for operational air-quality management in real-world applications.
title Interpretable Air Pollution Forecasting by Physics-Guided Spatiotemporal Decoupling
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2511.20257