遇见数据集

Structural Concentration of Extreme Precipitation in High-Ω States (Tokyo) — Minimal Reproducible Test

收藏
Zenodo2026-05-04 更新2026-05-26 收录
官方服务:

资源简介:

This record provides a minimal, reproducible test of structural concentration in a weather domain. Using daily precipitation data (Open-Meteo, Tokyo), we test whether extreme precipitation events concentrate in a predefined high-Ω regime under fixed, ex-ante definitions. Result: P(collapse | high Ω): 0.48 Baseline: 0.050 Ratio: 9.47× Ω is defined as: Ω = I × G I: rolling standard deviation of precipitation (window = 30) G: absolute daily change in precipitation Definitions: - high Ω = Ω > q(0.99) (execution layer) - collapse = precipitation > q(0.95) All definitions are fixed prior to evaluation and are independent (non-circular). This is not a predictive model. No optimization is performed. The result reports conditional frequency, not timing prediction. Reproducibility: https://colab.research.google.com/drive/1hCA98FhUjrJd9xYC9c02qm6pjcGpWhWP Data: Open-Meteo (public) Code: included in the Colab notebook

本数据集针对气象领域的结构集中度开展了一项极简可复现的测试。 本研究采用Open-Meteo提供的东京逐日降水数据,基于事前固定的判据,检验极端降水事件是否会集中于预先定义的高Ω状态区间内。 测试结果: P(极端降水 | 高Ω状态):0.48 基准值:0.050 比值:9.47倍 Ω的定义如下: Ω = I × G 其中,I为降水的滚动标准差(窗口长度为30),G为降水的日绝对变化量。 判据定义: - 高Ω状态:Ω大于0.99分位数(执行阈值) - 极端降水事件:单日降水量超过0.95分位数 所有判据均在评估前预先固定,且相互独立,不存在循环论证问题。 本测试未构建预测模型,未执行任何优化操作,所得结果仅为条件频率,而非时序预测结果。 可复现性链接:https://colab.research.google.com/drive/1hCA98FhUjrJd9xYC9c02qm6pjcGpWhWP 数据来源:Open-Meteo(公开数据集) 代码已包含于上述Colab笔记本中

提供机构:
Zenodo
创建时间:
2026-05-04
二维码
社区交流群
二维码
科研交流群
商业服务