基于矩阵轮廓与集成学习的多地理模型自适应降雨预测
机器学习
2025-09-15 v2
摘要
越南降雨预测极具挑战性 due to its diverse climatic conditions and strong geographical variability across river basins, yet accurate and reliable forecasts are vital for flood management, hydropower operation, and disaster preparedness. 本文提出一种基于矩阵轮廓的加权集成(MPWE),一个具有动态切换框架,能够捕获多个地理模型预测之间的协变量依赖关系,同时融入冗余感知加权以平衡跨模型的贡献。我们使用越南八大流域的降雨预测评估了 MPWE,涵盖五个预报时程(1 小时以及 12、24、48、72 和 84 小时的累计降雨)。实验结果表明,MPWE 在地理模型和集成基准方法中持续实现更低的预测误差均值和标准差,展示了在不同流域和时程下的 improved accuracy 和稳定性。
引用
@article{arxiv.2509.08277,
title = {Adaptive Rainfall Forecasting from Multiple Geographical Models Using Matrix Profile and Ensemble Learning},
author = {Dung T. Tran and Huyen Ngoc Huyen and Hong Nguyen and Xuan-Vu Phan and Nam-Phong Nguyen},
journal= {arXiv preprint arXiv:2509.08277},
year = {2025}
}