McCast: Memory-Guided Latent Drift Correction for Long-Horizon Precipitation Nowcasting
Penghui Wen, Yu Luo, Lintao Wang, Mengwei He, Patrick Filippi, Thomas Francis Bishop, Zhiyong Wang · 2026
Read McCast: Memory-Guided Latent Drift Correction for Long-Horizon Precipitation Nowcasting by Penghui Wen, Yu Luo, Lintao Wang, Mengwei He, Patrick Filippi, Thomas Francis Bishop, Zhiyong Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-13197.txt.
Introduction
Wen and six co-authors present McCast, a memory-guided method for correcting drift in long-horizon precipitation nowcasting. Autoregressive nowcasting predicts states from past outputs, accumulating errors that drag forecasts off course. Rather than treating memory as a loose pile of states, McCast uses time-ordered memory to correct drift. A Drift-Corrective Memory Bank estimates corrections in two stages: an extractor predicts a correction, then a retrieval module refines it from historical memory. On SEVIR and MeteoNet it reaches state-of-the-art performance over long horizons.
Catalog record
| CREATOR | Penghui Wen, Yu Luo, Lintao Wang, Mengwei He, Patrick Filippi, Thomas Francis Bishop, Zhiyong Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 90714 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.13197
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
McCast: Memory-Guided Latent Drift Correction for Long-Horizon Precipitation Nowcasting Penghui Wen, Yu Luo, Lintao Wang, Mengwei He, Patrick Filippi, Thomas Francis Bishop, Zhiyong Wang Abstract Existing precipitation nowcasting methods typically adopt an autoregressive formulation, where future states are predicted from previous outputs. However, such an approach accumulates errors over long rollouts, causing forecasts to drift away from physically plausible evolution trajectories. Although various studies have attempted to alleviate this problem by improving step-wise prediction accuracy, they largely neglect the global temporal evolution of meteorological systems and lack mechanisms to actively correct drift during rollouts. To address this issue, we propose McCast, a memory-guided latent drift correction method for precipitation nowcasting. Rather than treating memory as an unordered dictionary of latent states for passive conditioning, McCast leverages temporally organized memory to actively correct autoregressive latent evolution. Specifically, McCast introduces a Drift-Corrective Memory Bank (DCBank) that explicitly estimates the temporally consistent drift corrections to calibrate the divergent trajectory. DCBank performs drift correction in two stages: a Corrective Latent Extractor first predicts an initial correction from the current prediction and a reference latent state, and a Correction-Aware Memory Retrieval module then refines the initial correction using temporally organized historical memory. By explicitly correcting latent evolution, instead of improving step-wise prediction accuracy only, McCast produces more temporally coherent and reliable long-horizon forecasts. Experiments on two widely used benchmarks, SEVIR and MeteoNet, show that McCast achieves state-of-the-art performance, particularly in challenging long-horizon forecasting scenarios. 1 1 1 Code will be made publicly available upon acceptance. 1 Introduction Precipitation nowcasting aims to forecast near-future weather evolution from recent remote-sensing observations, such as weather radar, and plays a critical role in many aspects of society, such as transportation and disaster management [ 21 , 35 ] . Recently, deep learning-based approaches [ 33 , 17 ] have advanced precipitation nowcasting with strong performance and efficient inference, commonly framing it as an autoregressive process that predicts future frames step by step from previous outputs [ 21 , 10 ] . However, this formulation ignores the continuous temporal evolution of precipitation systems and treats each prediction step as locally optimal, causing the predicted sequence to progressively deviate from physically plausible evolution trajectories. Such compounding error, referred as precipitation drift (Figure 1 ) which manifests as degraded spatial structure, motion inconsistency, and loss of coherence in long-horizon forecasts [ 32 ] . Existing methods mainly focus on mitigating the precipitation drift by improving per-step accuracy through larger models or more sophisticated architectures [ 5 , 28 ] . However, due to the inherent stochasticity of atmospheric system, single-step prediction errors are inevitable, leaving the fundamental error accumulation problem unsolved. It motivates a paradigm shift from relying on locally optimal predictions to explicitly model global temporal evolution. A promising alternative is to introduce external memory [ 36 ] , which utilizes historical states as valuable cues about how precipitation intensity and motion evolve over time. In principle, such states can serve as a temporal anchor for long-horizon forecasting. However, existing vanilla memory-based designs [ 30 , 34 ] generally treat memory as a static and unordered dictionary of latent states for passive conditioning. Although this design allows the model to access historical context, simply using relevant historical content fails to model the temporal dependencies among memory states, and therefore cannot effectively capture the continuity of atmospheric evolution, nor can it provide targeted correction signals for precipitation drift. Therefore, as illustrated in Figure 1 , passive memory still exhibit noticeable drift, and the corresponding retrieved memories lack clear temporal consistency in both motion and intensity. Figure 1: Visualization of DiffCast [ 33 ] with a vanilla memory bank, including the prediction and retrieved memories. Background color denotes intensity, red arrows indicate motion, and the dotted box highlights a zoomed-in region. The prediction shows drift and the retrieved memory lacks temporal consistency in intensity and motion. To address this gap, we propose McCast , a memory-guided latent drift correction method for precipitation nowcasting that explicitly models temporal dependencies among historical memory and reformulates memory from passive conditioning into an active correction mechanism for autoregressive latent evolution. Specifically, McCast introduces a Drift-Corrective Memory Bank (DCBank) that explicitly models the temporal dependencies among historical states and uses them to estimate a drift correction term for the current latent prediction, and updates the prior latent state into a posterior latent state through explicit correction rather than implicit feature fusion. DCBank contains two components: a Corrective Latent Extractor , which produces an initial correction by characterizing the discrepancy between the current prediction and a reference latent state, and a Correction-Aware Memory Retrieval , which further refines the initial correction by selectively retrieving temporally organized memory states based on both semantic relevance and drift consistency. In this way, McCast explicitly enforces a temporally consistent evolution and reduces long-horizon precipitation drift. Our main contributions are summarized as follows: • We propose McCast, a memory-guided latent drift correction method that explicitly leverages temporal dependencies in memory to mitigate precipitation drift. • We introduce a Drift-Corrective Memory Bank (DCBank) that formulates memory as an explicit correction mechanism rather than a passive conditioning signal, enabling effective calibration of latent representations using historical evolution patterns. • We design a two-step drift correction scheme consisting of a Corrective Latent Extractor for initial drift correction estimation and a Correction-Aware Memory Retrieval, to produce refined correction signals for latent evolution. • Extensive experiments on two real-world precipitation datasets, SEVIR [ 26 ] and MeteoNet [ 16 ] , demonstrate that McCast achieves superior performance and robustness compared to existing methods. 2 Related Works Precipitation Nowcasting Models typically forecast future frames autoregressively by conditioning on previous outputs. Existing expert models can be broadly divided into deterministic, probabilistic, and hybrid approaches. Deterministic models focus on learning mean spatiotemporal evolution [ 24 ] , such as AlphaPre [ 17 ] , which decouples motion and intensity through phase-amplitude factorization, and PercpCast [ 5 ] , which improves perceptual quality via rectified flow. Probabilistic models introduce latent variables to capture forecast uncertainty and fine-scale structure [ 21 ] , as exemplified by Prediff [ 8 ] and DuoCast [ 28 ] . Hybrid methods, such as CasCast [ 10 ] and DiffCast [ 33 ] , combine deterministic trajectory prediction with stochastic refinement. Meanwhile, recent weather foundation models have shown strong transferability across forecasting tasks. For example, Aurora [ 1 ] has been adapted to diverse downstream applications such as cyclone tracks and weather prediction. Despite their effectiveness, these autoregressive approaches suffer from error accumulate over long rollouts, as informative historical states are gradually lost once they fall outside the short prediction window. To address it, we introduce a memory bank mechanism that preserves and retrieves historical states, providing a stable temporal anchor for long-horizon prediction. Memory Mechanisms in Video Generation have been widely adopted to preserve long-horizon consistency. Explicit memory methods, such as WorldExplorer [ 22 ] and Memory Forcing [ 15 ] , represent video content as 3D point clouds and use reprojection constraints to guide generation. While effective for reducing spatial drift, they are computationally costly and rely on robust 3D reconstruction pipelines [ 27 ] . In contrast, implicit latent memory methods store historical information in