AsyncCouple-Flow improves forecasts from mixed and missing data sources
AsyncCouple-Flow: Asynchronous Cross-Modal Coupling and Flow Matching for Spatio-Temporal Forecasting
Machine Learning
Summary
Forecasting things like weather or traffic often uses different types of data collected at different times and speeds, which creates problems when trying to combine them. The authors present AsyncCouple-Flow, a method that smartly picks important data points, links them across time and source using a flexible graph, and predicts future values without making errors build up over time. Their approach works well even when some data sources are missing and beats existing top methods on weather and traffic tests.
What this means in practice
- •For weather forecasters: Combine diverse slow and fast weather data streams to produce more accurate short-term forecasts even with sensor gaps or missing data.
- •For traffic management teams: Improve traffic flow predictions by integrating mixed data types with irregular timing and handle sensor outages robustly.
Authors
Zhixiang Wu, Yining Liu, Bo Zhao, Szu-Yu Chen, Huiran Duan, Chu Lin, Chuanguang Yang
Abstract
Multi-modal spatio-temporal forecasting (MM-STF) supports weather nowcasting, traffic prediction, and earth-system modeling by combining heterogeneous sources such as physical fields, satellite imagery, and in-situ sensors. Three obstacles persist: (i) modalities have different spatio-temporal sampling rates, forcing lossy interpolation onto a unified grid; (ii) modalities are frequently missing at deployment due to sensor outages or revisit gaps, while most methods train with full availability; and (iii) autoregressive decoders accumulate errors over long horizons, amplified by multi-modal conditioning. We propose AsyncCouple-Flow to address these issues jointly. A Modality-Aware Token Sparsification (MATS) module performs scale-aware tokenization and uses a shared importance scorer to select top-k tokens per timestep, producing equal-length sequences. An Asynchronous Cross-Modal Coupling Graph (ACCG) replaces fixed cross-attention with a learnable graph whose edges encode time offsets, semantic similarity, and modality-specific physical priors, enabling fusion under arbitrary asynchrony and missingness. A Flow-Matching Forecasting Head models multi-step prediction as a conditional ODE, trained with stochastic modality dropout and integrated jointly to avoid autoregressive drift. Experiments on ERA5+GOES+ISD weather forecasting and PEMS-BAY traffic prediction with multi-source side information show that AsyncCouple-Flow outperforms state-of-the-art baselines and remains robust with up to two missing modalities. The code will be released upon acceptance.