From https://metaculus.com//questions/15536/6-month-ai-weather-forecasting/ Advancements in artificial intelligence (AI) and machine learning have led to significant improvements in many fields, [including weather forecasting](https://source.colostate.edu/ai-and-machine-learning-are-improving-weather-forecasts-but-they-wont-replace-human-experts/). Traditional weather forecasting models rely on complex simulations and data from numerous sources, such as satellites, weather stations, and ocean buoys. However, these models typically have [limited accuracy beyond a week or two](https://en.wikipedia.org/wiki/Numerical_weather_prediction), making long-term predictions challenging; weather forecasts beyond 10 days are only right [half the time](https://scijinks.gov/forecast-reliability/#:~:text=A%20seven%2Dday%20forecast%20can,right%20about%20half%20the%20time.). As climate change continues to disrupt weather patterns, the need for more accurate and longer-term forecasts has become increasingly important for planning purposes and mitigating the adverse effects of extreme weather events. In recent years, AI techniques, such as deep learning, have shown [promising results](https://eos.org/research-spotlights/the-ai-forecaster-machine-learning-takes-on-weather-prediction) in enhancing short-term weather predictions. Given this potential, it is of great interest to determine whether AI can be utilized to significantly extend the time horizon for accurate weather forecasts. A breakthrough in long-term forecasting could transform industries like agriculture, renewable energy, and disaster management, helping society adapt to climate change more effectively. ***Will an AI model be developed before 2030 that can accurately predict local weather patterns up to 6 months in advance?*** This question will be considered resolved as **Yes** if, by December 31, 2029, a peer-reviewed study or an official announcement from a recognized meteorological organization confirms the development of an AI model that: 1. Predicts local weather patterns (e.g., temperature, precipitation) with a lead time of up to 6 months. 2. Achieves accuracy levels which surpass the [2022 state-of-the-art for seven-day forecasts](https://www.wpc.ncep.noaa.gov/html/hpcverif.shtml), defined as: - for precipitation, a "[threat score](https://www.e-education.psu.edu/meteo3/node/2285#:~:text=Threat%20scores%20indicate%20that%20the%20Weather%20Prediction%20Center%27s,less%20than%20half%20the%20area%20correct%2C%20on%20average.)" for 1-inch precipitation of at least [0.18](https://www.wpc.ncep.noaa.gov/images/hpcvrf/wpcd4710yr.gif), and; - for temperature, a [Mean Absolute Error](https://learningweather.psu.edu/node/77#:~:text=Mean%20Absolute%20Error,-Perhaps%20the%20simplest&text=So%2C%20if%20you%20forecast%20a,%3D%20%7C%2D3%C2%BAF%7C%20%3D%203%C2%BAF.) of no worse than [4 degrees Fahrenheit](https://www.wpc.ncep.noaa.gov/images/hpcvrf/d7minyr.gif) for the minimum temperature and [4.7 degrees Fahrenheit](https://www.wpc.ncep.noaa.gov/images/hpcvrf/d7maxyr.gif) for the maximum temperature. 3. Is tested on real-world data across diverse climate zones. If the AI model is a component of a larger AI system, the question resolves **Yes** as long as the above criteria are met.