针对暴雨预测数据不平衡问题，提出增强型模型提升预警精度。
DPSformer: A long-tail-aware model for improving heavy rainfall prediction
- 设计多分支结构，重点增强强降雨事件的特征表示。
- 强降雨（≥50mm/6h）CSI指标从0.012提升至0.067。
- 适用于气象预警系统，尤其适合极端天气研究者。
准确及时地预测强降雨仍是现代社会的重大挑战。降水分布高度不均衡：多数观测为无雨或小雨，而强降雨事件极为罕见。这种不平衡阻碍了深度学习模型对强降雨的有效预测。为此，我们将降雨预测明确视为长尾学习问题，识别出强降雨事件表征不足是精度瓶颈。因此，我们提出DPSformer，一种长尾感知模型，通过高分辨率分支增强强降雨事件的表征。对于≥50 mm/6 h的强降雨事件，该模型将基准数值天气预报（NWP）模型的临界成功指数（CSI）从0.012提升至0.067；在最强降雨事件前1%覆盖率下，分数技能得分（FSS）超过0.45，优于现有方法。本工作建立了一种有效的长尾范式，为提升早期预警系统、减轻极端天气社会影响提供了实用工具。
原文摘要 · Abstract (English)
Accurate and timely forecasting of heavy rainfall remains a critical challenge for modern society. Precipitation exhibits a highly imbalanced distribution: most observations record no or light rain, while heavy rainfall events are rare. Such an imbalanced distribution obstructs deep learning models from effectively predicting heavy rainfall events. To address this challenge, we treat rainfall forecasting explicitly as a long-tailed learning problem, identifying the insufficient representation of heavy rainfall events as the primary barrier to forecasting accuracy. Therefore, we introduce DPSformer, a long-tail-aware model that enriches representation of heavy rainfall events through a high-resolution branch. For heavy rainfall events $ \geq $ 50 mm/6 h, DPSformer lifts the Critical Success Index (CSI) of a baseline Numerical Weather Prediction (NWP) model from 0.012 to 0.067. For the top 1% coverage of heavy rainfall events, its Fraction Skill Score (FSS) exceeds 0.45, surpassing existing methods. 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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。