E-STGCN: Extreme Spatiotemporal Graph Convolutional Networks for Air Quality Forecasting

2024-11-19

AI Summary

To address the inadequate modeling of extreme events in air quality forecasting for highly polluted cities (e.g., Delhi), this paper proposes the first probabilistic forecasting framework integrating Extreme Value Theory—specifically the Generalized Pareto Distribution (GPD)—with Spatio-Temporal Graph Convolutional Networks (STGCN), augmented by conformal prediction to yield statistically rigorous uncertainty intervals. The method explicitly captures the nonlinear, non-stationary, and spatio-temporal extremal dependencies of PM₂.₅, PM₁₀, and NO₂ concentrations, overcoming the tail-risk modeling limitations of conventional time-series and spatio-temporal models. Evaluated on full-season data from 37 monitoring stations across Delhi, the framework significantly outperforms state-of-the-art temporal and spatio-temporal baselines, particularly in identifying exceedance risks. The resulting prediction intervals provide valid statistical coverage guarantees and intrinsic interpretability, establishing a novel paradigm for robust decision-making under severe pollution conditions.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Calibration & Uncertainty Quantification
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Modeling and forecasting air quality plays a crucial role in informed air pollution management and protecting public health. The air quality data of a region, collected through various pollution monitoring stations, display nonlinearity, nonstationarity, and highly dynamic nature and detain intense stochastic spatiotemporal correlation. Geometric deep learning models such as Spatiotemporal Graph Convolutional Networks (STGCN) can capture spatial dependence while forecasting temporal time series data for different sensor locations. Another key characteristic often ignored by these models is the presence of extreme observations in the air pollutant levels for severely polluted cities worldwide. Extreme value theory is a commonly used statistical method to predict the expected number of violations of the National Ambient Air Quality Standards for air pollutant concentration levels. This study develops an extreme value theory-based STGCN model (E-STGCN) for air pollution data to incorporate extreme behavior across pollutant concentrations. Along with spatial and temporal components, E-STGCN uses generalized Pareto distribution to investigate the extreme behavior of different air pollutants and incorporate it inside graph convolutional networks. The proposal is then applied to analyze air pollution data (PM2.5, PM10, and NO2) of 37 monitoring stations across Delhi, India. The forecasting performance for different test horizons is evaluated compared to benchmark forecasters (both temporal and spatiotemporal). It was found that E-STGCN has consistent performance across all the seasons in Delhi, India, and the robustness of our results has also been evaluated empirically. Moreover, combined with conformal prediction, E-STGCN can also produce probabilistic prediction intervals.

Problem

• Modeling extreme air pollutant levels in spatiotemporal data
• Incorporating extreme value theory into deep learning for air quality
• Improving forecasting accuracy for severely polluted cities like Delhi

Innovation

• EVT-guided modified STGCN model
• Graph convolutional networks for spatial modeling
• Generalized Pareto distribution for extreme behavior