Scalable non-separable spatio-temporal Gaussian process models for large-scale short-term weather prediction

2026-02-03

AI Summary

This study addresses the computational challenges of applying large-scale spatiotemporal Gaussian processes to continental-scale short-term weather forecasting. Focusing on daily maximum temperature and precipitation prediction over the contiguous United States, the authors propose a scalable nonseparable spatiotemporal Gaussian process model that integrates FITC, Vecchia approximation, and a VIF-based hybrid strategy. The approach incorporates correlation-based neighborhood selection and a spatiotemporal kMeans++ algorithm for inducing point placement. Key computational components are accelerated via GPU implementation, substantially improving efficiency. Experiments on approximately 1.7 million observations demonstrate that the method outperforms existing approaches in predictive accuracy, parameter estimation, and computational speed, marking the first successful realization of efficient and accurate continental-scale Gaussian process–based weather forecasting.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Machine Learning: Kernel Methods
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Monitoring daily weather fields is critical for climate science, agriculture, and environmental planning, yet fully probabilistic spatio-temporal models become computationally prohibitive at continental scale. We present a case study on short-term forecasting of daily maximum temperature and precipitation across the conterminous United States using novel scalable spatio-temporal Gaussian process methodology. Building on three approximation families - inducing-point methods (FITC), Vecchia approximations, and a hybrid Vecchia-inducing-point full-scale approach (VIF) - we introduce three extensions that address key bottlenecks in large space-time settings: (i) a scalable correlation-based neighbor selection strategy for Vecchia approximations with point-referenced data, enabling accurate conditioning under complex dependence structures, (ii) a space-time kMeans++ inducing-point selection algorithm, and (iii) GPU-accelerated implementations of computationally expensive operations, including matrix operations and neighbor searches. Using both synthetic experiments and a large NOAA station dataset containing more than one million space-time observations, we analyze the models with respect to predictive performance, parameter estimation, and computational efficiency. Our results demonstrate that scalable Gaussian process models can yield accurate continental-scale forecasts while remaining computationally feasible, offering practical tools for weather applications.

Problem

• spatio-temporal Gaussian process
• large-scale weather prediction
• computational scalability
• short-term forecasting
• continental-scale modeling

Innovation

• scalable Gaussian processes
• Vecchia approximation
• inducing-point methods
• space-time kMeans++
• GPU acceleration