基于概率密度标签的自监督学习用于降雨概率估计
机器学习
2024-12-10 v1 计算机视觉与模式识别
摘要
数值天气预报（NWP）模型是气象学中模拟和预测各种大气变量的基础。降雨预测的准确性以及获取足够预报时程对于防止有害天气事件至关重要。然而，NWP模型的性能受限于由时序动力学驱动的极端天气现象的非线性和不可预测模式。就此而言，我们提出了一种基于概率密度标签的自监督学习方法（SSLPDL）用于估计降雨概率。我们的后处理方法使用带掩码建模的自监督学习（SSL）来重建大气物理变量，使模型能够学习变量之间的依赖关系。预训练的编码器随后用于迁移学习到降雨分割任务。Furthermore, we introduce a straightforward labeling approach based on probability density to address the class imbalance in extreme weather phenomena like heavy rain events. Experimental results show that SSLPDL surpasses other precipitation forecasting models in regional precipitation post-processing and demonstrates competitive performance in extending forecast lead times. Our code is available at https://github.com/joonha425/SSLPDL
引用
@article{arxiv.2412.05825,
title = {Self-Supervised Learning with Probabilistic Density Labeling for Rainfall Probability Estimation},
author = {Junha Lee and Sojung An and Sujeong You and Namik Cho},
journal= {arXiv preprint arXiv:2412.05825},
year = {2024}
}
备注
Accepted by WACV 2025