Published October 18, 2026
| Version 2026.10
Working paper
Open
Weather in the News, 2019–2026: Seasonality, Geography and Tone of Global Weather Coverage, and How It Compares with Research and Invention
Description
We measure how the weather was covered in a global news index from January 2019 to July 2026, using 480,612 de-duplicated weather articles, and set that coverage beside 35,955 research papers (2015–2025) and 9,541 patent families (2015–2024) on weather and forecasting. Six results stand out:
- Seasonality. January carries the largest share of each year's weather coverage (10.63%), and December to February together carry 28.35%. A second, smaller peak falls in July.
- Attention, not weather. Weather took a 33% smaller share of news in 2020 than in 2019. Its share peaked in January 2026, the month of a large North American winter storm, at 54.57 articles per 10,000.
- Geography. India's share of located weather news is 2.43 times its share of all news. The United States, the largest source of located weather articles (39.23%), is under-represented at 0.81. Sri Lanka (3.38), India and New Zealand (2.2) are the most over-represented of the leading countries. Within India, Himachal Pradesh (3.76), Odisha (1.93) and Sikkim (1.93) lead; within the United States, Mississippi (4.03), Nebraska (3.2), North Dakota and New Mexico (both 2.66).
- Tone and urgency. 63.9% of weather articles are negative in tone (all news: 33.6%), and 53.4% are rated critical priority (all news: 42.6%).
- Measurement. 29.36% of raw matches for weather vocabulary were automated warning bulletins, machine-generated forecasts, page furniture or figurative uses. Counting them would inflate coverage.
- Research and invention. Weather papers grew from 4.52 to 6.42 per 10,000 papers from 2015 to 2025. Weather-forecasting patent families rose from 1.05 to 2.95 per 10,000 patent families from 2015 to 2024 (2.8-fold), and 86.0% of office filings were in China.
Notes
Files
FINDINGS.md
Additional details
Related works
- Is supplement to
- Software: https://github.com/fliarbi/prism-news-api (URL)
- Is supplemented by
- Dataset: 10.5281/zenodo.23445854 (DOI)
- Is variant form of
- Dataset: https://huggingface.co/datasets/fliarbi/weather-news-2019-2026 (URL)
- Dataset: https://www.kaggle.com/datasets/fliarbi/weather-news-dataset (URL)