Tailweather dataset helps improve rare extreme weather forecasts worldwide
TailWeather: from tail to extremes, a global climatological dataset for machine-learning weather forecasting
Computational Engineering, Finance, and Science
Summary
Weather forecasts usually work well on average but often miss rare and extreme events like heatwaves or heavy storms. The authors created TailWeather, a detailed global dataset that identifies extreme weather events daily and monthly using consistent standards from past climate data. This helps researchers and forecasters better detect and study dangerous weather extremes, beyond average conditions. It also shows that traditional disaster records miss many significant weather events.
What this means in practice
- •For weather forecast modelers: Use TailWeather as a reliable reference dataset to improve detection and evaluation of extreme weather in machine learning forecast models.
- •For insurance risk analysts: Incorporate TailWeather data to better understand regional weather extremes and improve models for catastrophe insurance and risk assessment.
Authors
Zhi-Song Liu, Michael Boy, Risto Makkonen
Abstract
Weather forecasting models commonly use the average forecast skill for short-range forecasts. However, it does not necessarily imply skill in the tails of the weather distribution. Evaluating tail events requires a consistently defined target with broad spatial and temporal coverage. Disaster catalogs record societal consequences, but are sparse and reporting-dependent; climatological tails describe unusual weather without necessarily implying harm. We present TailWeather, a global 0.25-degree, land-only dataset derived from ERA5, covering 1981-2022 and extending into January 2023. It labels heatwaves, cold waves, heavy precipitation, and extreme wind daily, and meteorological drought monthly. Each event has an ordinal severity tier and a numerical intensity score referenced to the local 1991-2020 climate. The scores support alternative thresholds within their stored resolution and valid domain. Comparison with documented disasters shows greater impact enrichment towards stricter tails, with differences among hazards and substantial gaps in catalog coverage. Forecast examples illustrate how low average errors can coexist with weak event detection, particularly for wind. TailWeather provides a reusable physical target for studying and evaluating extremes, while complementing the information in disaster catalogs.