Seeing Inside the Storm: Improving Nowcasting by Integrating Meteorological Drivers

2026-05-22

AI Summary

This study addresses the limitations of current nowcasting systems, which predominantly rely on radar echoes and struggle to detect subtle precursors of convective storm initiation—such as low-level convergence, turbulent vortices, and latent heat release. To overcome this, the authors propose MeteoLogist, a novel framework that, for the first time, enables spatiotemporal co-modeling of asynchronously evolving multiscale meteorological drivers, including thermodynamic, dynamic, and microphysical factors. The approach integrates physics-informed encoders, a causal temporal attention mechanism for phase alignment, and a cross-field spatial aggregator to deeply fuse heterogeneous multimodal data. Evaluated on the 3D-NEXRAD dataset, MeteoLogist improves the CSI40 score for high-impact precipitation detection by 9.7% and enhances early-stage storm prediction capability by 37.67%, demonstrating significant advances in convective initiation forecasting.

Abstract

Most nowcasting systems, built on radar reflectivity, focus on current precipitation, ignoring the atmospheric precursors -- such as low-level convergence, turbulent eddies, and latent heating -- that offer a fleeting window to foresee storm birth. We introduce MeteoLogist, a physics-inspired radar intelligence framework that models the full life cycle of convection -- from its precursors to organized storm evolution. However, exploiting these precursors is non-trivial: they originate from multiple meteorological drivers -- thermodynamic, kinematic, and microphysical -- that evolve asynchronously (C1) and remain spatially fragmented (C2). To this end, MeteoLogist designs three tightly integrated components. The Physics-Tailored Encoders process radar echoes according to their intrinsic physical scales and semantics, forming thermodynamic, kinematic, and microphysical streams that capture distinct dynamical regimes. The Temporal-Phase Aligner addresses C1 by leveraging causal temporal attention to capture when and how different drivers interact and activate. The Cross-Field Spatial Aggregator addresses C2 through cross-regional fusion, aligning weak and scattered precursors across neighboring cells to expose upstream triggers and enforce spatial coherence. Evaluated on 3D-NEXRAD (2020--2022, US-wide), MeteoLogist boosts high-impact detection (CSI40) by +9.7% over strong baselines, and achieves a remarkable 37.67% gain during the storm-developing stage -- demonstrating true foresight in sensing storms before they appear. The code can be found in the supplementary material.

Problem

• nowcasting
• storm precursors
• meteorological drivers
• convection
• radar reflectivity

Innovation

• nowcasting
• meteorological drivers
• physics-inspired framework
• temporal-phase alignment
• spatial aggregation