Interpretable long-horizon air pollution forecasting using a transformer-based framework

doi:10.1016/j.eswa.2026.131719

Interpretable long-horizon air pollution forecasting using a transformer-based framework

2026-02-18

0

OA

AI

DOI：10.1016/j.eswa.2026.131719

Original

Save

View PDF

Abstract

En 中文

• Integrates heterogeneous forecasted variables to enhance long-horizon forecasting performance. • Provides fine-grained interpretability for each forecasted timestep across temporal and feature dimensions. • Achieves superior predictive accuracy for NOX and PM10 across forecast horizons up to 720 hours.

Keywords:

Multi-horizon forecasting

Explainable AI

Meteorological forecast integration

Air quality and health risk

Interpretable long-horizon air pollution forecasting using a transformer-based framework

Abstract

AI Summary

Journal

Organization