PA-Net: Precipitation-Adaptive Mixture-of-Experts for Long-Tail Rainfall Nowcasting

• URL: http://arxiv.org/abs/2603.13818v1
• Date: Sat, 14 Mar 2026 08:02:07 GMT
• Title: PA-Net: Precipitation-Adaptive Mixture-of-Experts for Long-Tail Rainfall Nowcasting
• Abstract summary: Precipitation nowcasting is vital for flood warning, agricultural management, and emergency response.<n>We propose the Precipitation-Adaptive Network (PA-Net), a Transformer framework whose computational budget is explicitly governed by rainfall intensity.<n>Experiments on ERA5 demonstrate consistent improvements over state-of-the-art baselines, with significant gains in heavy-rain and rainstorm regimes.
• Score: 30.94380402656709
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Precipitation nowcasting is vital for flood warning, agricultural management, and emergency response, yet two bottlenecks persist: the prohibitive cost of modeling million-scale spatiotemporal tokens from multi-variate atmospheric fields, and the extreme long-tailed rainfall distribution where heavy-to-torrential events -- those of greatest societal impact -- constitute fewer than 0.1% of all samples. We propose the Precipitation-Adaptive Network (PA-Net), a Transformer framework whose computational budget is explicitly governed by rainfall intensity. Its core component, Precipitation-Adaptive MoE (PA-MoE), dynamically scales the number of activated experts per token according to local precipitation magnitude, channeling richer representational capacity toward the rare yet critical heavy-rainfall tail. A Dual-Axis Compressed Latent Attention mechanism factorizes spatiotemporal attention with convolutional reduction to manage massive context lengths, while an intensity-aware training protocol progressively amplifies learning signals from extreme-rainfall samples. Experiment on ERA5 demonstrate consistent improvements over state-of-the-art baselines, with particularly significant gains in heavy-rain and rainstorm regimes.

Related papers

• MZ-Rain: Moisture-Budget-Guided Zero-Inflated Model for Station-Level Precipitation Nowcasting [18.922294934481464]
  We propose MZ-Rain, a moisture-budget-guided zero-inflated sLSTM framework for station-level precipitation nowcasting.<n>MZ-Rain decomposes the precipitation formation process into process-specific pathways corresponding to moisture storage, moisture transport, surface evaporation, and precipitation persistence.<n>It consistently outperforms strong baselines on multiple evaluation metrics, including CSI, FAR, MSE, and MAE.
  arXiv  Detail & Related papers  (2026-09-04T08:24:52Z)
• QWRF-Net: A Quantum-Wavelet Framework with Rectified Flow for Short-Term Precipitation Nowcasting [4.271790881899781]
  Key challenge in warning-oriented nowcasting is that radar precipitation fields contain strongly coupled multi-scale structures.<n>We propose QWRF-Net, a quantum-wavelet framework with rectified flow for short-term precipitation nowcasting.
  arXiv  Detail & Related papers  (2026-08-03T02:58:49Z)
• MP-MoE: Matrix Profile-Guided Mixture of Experts for Precipitation Forecasting [0.9786690381850356]
  We propose the Matrix Profile-guided Mixture of Experts (MP-MoE) framework.<n>MP-MoE integrates conventional intensity loss with a structural-aware Matrix Profile objective.<n>We evaluate MP-MoE on rainfall datasets from two major river basins in Vietnam.
  arXiv  Detail & Related papers  (2026-03-26T05:29:17Z)
• CASCADE: Cross-scale Advective Super-resolution with Climate Assimilation and Downscaling Evolution [51.56484100374058]
  CASCADE reconstructs fine structure by iteratively advecting coarse information along learned, flow-conditioned velocity warping fields.<n> evaluated on SEVIR radar data for 4x super-resolution of severe convective storms.<n> produces temporally coherent, physically consistent, and mass-conserving reconstructions well suited to advection-dominated extremes in atmospheric and oceanic applications.
  arXiv  Detail & Related papers  (2026-02-18T14:41:54Z)
• Extreme Weather Nowcasting via Local Precipitation Pattern Prediction [6.992919908851609]
  ExPreCast is an efficient deterministic framework for generating detailed radar forecasts.<n>Our approach achieves state-of-the-art performance, delivering accurate and reliable nowcasts across both normal and extreme rainfall regimes.
  arXiv  Detail & Related papers  (2026-02-05T01:55:14Z)
• SynCast: Synergizing Contradictions in Precipitation Nowcasting via Diffusion Sequential Preference Optimization [62.958457694151384]
  We introduce preference optimization into precipitation nowcasting for the first time, motivated by the success of reinforcement learning from human feedback in large language models.<n>In the first stage, the framework focuses on reducing FAR, training the model to effectively suppress false alarms.
  arXiv  Detail & Related papers  (2025-10-22T16:11:22Z)
• DPSformer: A long-tail-aware model for improving heavy rainfall prediction [48.884870685632755]
  We introduce DPSformer, a long-tail-aware model that enriches representation of heavy rainfall events through a high-resolution branch.<n>For heavy rainfall events, DPSformer lifts the Critical Success Index (CSI) of a baseline Numerical Weather Prediction (NWP) model from 0.012 to 0.067.<n>Our work establishes an effective long-tailed paradigm for heavy rainfall prediction, offering a practical tool to enhance early warning systems and mitigate the societal impacts of extreme weather events.
  arXiv  Detail & Related papers  (2025-09-20T15:09:38Z)
• TRG-Net: An Interpretable and Controllable Rain Generator [61.2760968459789]
  This study proposes a novel deep learning based rain generator, which fully takes the physical generation mechanism underlying rains into consideration. Its significance lies in that the generator not only elaborately design essential elements of the rain to simulate expected rains, but also finely adapt to complicated and diverse practical rainy images. Our unpaired generation experiments demonstrate that the rain generated by the proposed rain generator is not only of higher quality, but also more effective for deraining and downstream tasks.
  arXiv  Detail & Related papers  (2024-03-15T03:27:39Z)
• CasCast: Skillful High-resolution Precipitation Nowcasting via Cascaded Modelling [93.65319031345197]
  We propose CasCast, a cascaded framework composed of a deterministic and a probabilistic part to decouple predictions for mesoscale precipitation distributions and small-scale patterns. CasCast significantly surpasses the baseline (up to +91.8%) for regional extreme-precipitation nowcasting.
  arXiv  Detail & Related papers  (2024-02-06T08:30:47Z)
• Residual Corrective Diffusion Modeling for Km-scale Atmospheric Downscaling [58.456404022536425]
  State of the art for physical hazard prediction from weather and climate requires expensive km-scale numerical simulations driven by coarser resolution global inputs. Here, a generative diffusion architecture is explored for downscaling such global inputs to km-scale, as a cost-effective machine learning alternative. The model is trained to predict 2km data from a regional weather model over Taiwan, conditioned on a 25km global reanalysis.
  arXiv  Detail & Related papers  (2023-09-24T19:57:22Z)
• TRU-NET: A Deep Learning Approach to High Resolution Prediction of Rainfall [21.399707529966474]
  We present TRU-NET, an encoder-decoder model featuring a novel 2D cross attention mechanism between contiguous convolutional-recurrent layers. We use a conditional-continuous loss function to capture the zero-skewed %extreme event patterns of rainfall. Experiments show that our model consistently attains lower RMSE and MAE scores than a DL model prevalent in short term precipitation prediction.
  arXiv  Detail & Related papers  (2020-08-20T17:27:59Z)
• From Rain Generation to Rain Removal [67.71728610434698]
  We build a full Bayesian generative model for rainy image where the rain layer is parameterized as a generator. We employ the variational inference framework to approximate the expected statistical distribution of rainy image. Comprehensive experiments substantiate that the proposed model can faithfully extract the complex rain distribution.
  arXiv  Detail & Related papers  (2020-08-08T18:56:51Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.