Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hybrid Quantum-Classical Spatiotemporal Forecasting for 3D Cloud Fields

Fu Wang, Qifeng Lu, Xinyu Long, Meng Zhang, Xiaofei Yang, Weijia Cao, Xiaowen Chu

Abstract

Accurate forecasting of three-dimensional (3D) cloud fields is important for atmospheric analysis and short-range numerical weather prediction, yet it remains challenging because cloud evolution involves cross-layer interactions, nonlocal dependencies, and multiscale spatiotemporal dynamics. Existing spatiotemporal prediction models based on convolutions, recurrence, or attention often rely on locality-biased representations and therefore struggle to preserve fine cloud structures in volumetric forecasting tasks. To address this issue, we propose QENO, a hybrid quantum-inspired spatiotemporal forecasting framework for 3D cloud fields. The proposed architecture consists of four components: a classical spatiotemporal encoder for compact latent representation, a topology-aware quantum enhancement block for modeling nonlocal couplings in latent space, a dynamic fusion temporal unit for integrating measurement-derived quantum features with recurrent memory, and a decoder for reconstructing future cloud volumes. Experiments on CMA-MESO 3D cloud fields show that QENO consistently outperforms representative baselines, including ConvLSTM, PredRNN++, Earthformer, TAU, and SimVP variants, in terms of MSE, MAE, RMSE, SSIM, and threshold-based detection metrics. In particular, QENO achieves an MSE of 0.2038, an RMSE of 0.4514, and an SSIM of 0.6291, while also maintaining a compact parameter budget. These results indicate that topology-aware hybrid quantum-classical feature modeling is a promising direction for 3D cloud structure forecasting and atmospheric Earth observation data analysis.

Hybrid Quantum-Classical Spatiotemporal Forecasting for 3D Cloud Fields

Abstract

Accurate forecasting of three-dimensional (3D) cloud fields is important for atmospheric analysis and short-range numerical weather prediction, yet it remains challenging because cloud evolution involves cross-layer interactions, nonlocal dependencies, and multiscale spatiotemporal dynamics. Existing spatiotemporal prediction models based on convolutions, recurrence, or attention often rely on locality-biased representations and therefore struggle to preserve fine cloud structures in volumetric forecasting tasks. To address this issue, we propose QENO, a hybrid quantum-inspired spatiotemporal forecasting framework for 3D cloud fields. The proposed architecture consists of four components: a classical spatiotemporal encoder for compact latent representation, a topology-aware quantum enhancement block for modeling nonlocal couplings in latent space, a dynamic fusion temporal unit for integrating measurement-derived quantum features with recurrent memory, and a decoder for reconstructing future cloud volumes. Experiments on CMA-MESO 3D cloud fields show that QENO consistently outperforms representative baselines, including ConvLSTM, PredRNN++, Earthformer, TAU, and SimVP variants, in terms of MSE, MAE, RMSE, SSIM, and threshold-based detection metrics. In particular, QENO achieves an MSE of 0.2038, an RMSE of 0.4514, and an SSIM of 0.6291, while also maintaining a compact parameter budget. These results indicate that topology-aware hybrid quantum-classical feature modeling is a promising direction for 3D cloud structure forecasting and atmospheric Earth observation data analysis.

Paper Structure

This paper contains 21 sections, 10 equations, 4 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Classical Spatiotemporal Prediction Model
  4. RNN-based Methods
  5. CNN-Based Methods and attention mechanisms
  6. Quantum-enhanced Machine Learning Model
  7. The Proposed Quantum-Enhanced Method
  8. Classical Spatial-Temporal Encoder
  9. Topological Entanglement Quantum Enhancer
  10. Dynamic Fusion Temporal Unit
  11. Classical Spatial-Temporal Decoder
  12. Algorithmic Summary
  13. Experiment and Results
  14. Dataset Desciption and Baseline Models
  15. Analysis of Comparative Experiment
  16. ...and 6 more sections

Figures (4)

  • Figure 1: Comparison of 2D, 3D CNN and quantum-enhanced feature extraction.
  • Figure 2: The overall architecture of QENO
  • Figure 3: Visualization of 3D cloud forecasts for Channel 15 from the CMA‐MESO dataset. Top: Ground truth. Below: model predictions by QENO, TAU, Earthformer, PredRNN_Plus, ConvLSTM, SimVP_Plus, PhyDNet, MAU, and SimVP. This layout illustrates each model’s capability to reproduce fine‐scale cloud patterns and overall structural fidelity.
  • Figure 4: Quantum coherence matrix analysis: inter-feature correlation patterns across model variants. Panels (a)--(f) show pairwise correlation (coherence) of latent feature activations computed over the validation set for six model variants: (a) QENO_Full (full quantum-enhanced model), (b) QENO without the Dynamic Fusion Temporal Unit (DFTU disabled), (c) QENO with quantum layers removed from the decoder, (d) QENO with a lightweight MLP decoder, (e) QENO with a classical convolutional decoder, and (f) a purely classical LSTM baseline. Color encodes Pearson-style correlation (range $[-1,1]$); strong off-diagonal structure indicates pronounced non-local feature coherence. The full QENO model (a) exhibits large-scale block structure and prominent off-diagonal correlations, whereas ablated or classical variants (b)--(f) progressively lose these organized long-range patterns and revert toward local/diagonal correlation.