# Perché le previsioni del tempo sbagliano? | Giacomo Poletti | TEDxTrento Source: https://www.youtube.com/watch?v=F_PTJG6K2-E Recap page: https://rapidrecap.app/video/F_PTJG6K2-E Generated: 2025-12-03T18:33:15.494+00:00 --- ## Quick Overview Giacomo Poletti explains that weather forecast models often fail in complex terrain like the Alps because their coarse resolution (8km grid) smooths out crucial local features, leading to inaccuracies, especially regarding intense local phenomena like hailstorms, which are better predicted by high-resolution, limited-area models (1km grid). **Key Points:** - Poletti has 15 years of experience teaching meteorology at the Collegio delle Guide Alpine, giving him practical insight into forecasting errors. - A major issue in forecasting severe weather, like the hailstorm that hit Milan Marittima, is the coarse resolution (8km grid) of global models, which smooths out terrain features. - Local, high-resolution models (1km grid) capture topography better, as illustrated by the comparison between the global (8km) and limited-area (1km) orography maps. - The speaker recounts a personal experience where a global forecast failed to predict a severe local event, highlighting the danger of relying solely on coarse models. - Radiosondes are launched every 12 hours from 8 airports globally, providing crucial atmospheric data, but this data still requires sophisticated local modeling to be useful for highly localized events. - Poletti emphasizes that the future of accurate forecasting lies in embracing high-resolution, local models that can resolve fine-scale atmospheric dynamics, as opposed to relying on smoothed global outputs. ![Screenshot at 11:48: The speaker points to a slide comparing a global model map \(coarse grid, left\) with a high-resolution local model map \(detailed topography, right\), demonstrating how the global model smooths out essential mountain features, thus failing to capture localized severe weather risks accurately.](https://ss.rapidrecap.app/screens/F_PTJG6K2-E/00-11-48.png) **Context:** Giacomo Poletti, a meteorology instructor, addresses the common question of why weather forecasts, particularly concerning localized severe weather events like hailstorms, are frequently inaccurate. He uses the analogy of a hiker in a cave during a lightning storm to illustrate the danger of ignoring local details, contrasting the coarse resolution of global weather models with the precision required for accurate local predictions. ## Detailed Analysis Giacomo Poletti argues that weather forecast inaccuracies, especially concerning severe local phenomena, stem primarily from the resolution limitations of global weather models. He shares an anecdote about a dangerous storm in Brenta where a general forecast proved insufficient. Global models, like the one showing an 8km grid, simplify complex terrain features (like the Dolomites shown in a slide comparison), meaning they miss critical local effects. Local, limited-area models, using a 1km grid, provide much finer detail, accurately depicting topography and thus improving short-term severe weather prediction. He cites the example of a hailstorm in Milan Marittima where the coarse model failed to capture the intensity, while the high-resolution forecast was more accurate. Poletti notes that while satellite data and radiosonde launches provide vast amounts of atmospheric data, the interpretation and application of this data through high-resolution models are essential for predicting localized, life-threatening events accurately, contrasting this with the overly smooth outputs of global models. ### Introduction and Personal Experience - Poletti has 15 years teaching meteorology; recounts a scary experience where a local storm hit despite general forecasts being calm. ### The Problem of Model Resolution - Global models use an 8km grid, smoothing out topography (like the Dolomites), which leads to poor localized forecasts; a 1km resolution is required for accuracy. ### Visualizing the Difference - Compares the smoothed output of an 8km global model (red box) with the detailed output of a 1km local model (green box) for orography, showing the local model captures valleys and peaks correctly. ### Data Sources and Improvement - Data comes from global satellite observations and radiosondes launched hourly from 8 airports, but the way this data is processed locally determines forecast quality. ### Conclusion and Future Outlook - The future is in high-resolution modeling that captures fine-scale atmospheric details, moving beyond the simplistic view of global models to accurately predict localized, severe weather events. ![Screenshot at 00:01: Opening title card for TEDxTrento.](https://ss.rapidrecap.app/screens/F_PTJG6K2-E/00-00-01.png) ![Screenshot at 08:08: The speaker gestures towards the audience while holding the remote, emphasizing the scale of the problem.](https://ss.rapidrecap.app/screens/F_PTJG6K2-E/00-08-08.png) ![Screenshot at 11:48: Slide displaying two maps: the left \(red border\) shows the coarse 8km global model output, and the right \(green border\) shows the detailed 1km local model output for orography.](https://ss.rapidrecap.app/screens/F_PTJG6K2-E/00-11-48.png) ![Screenshot at 13:34: Detailed comparison slide showing the difference in topographical representation between the 8km global model \(left, smoothed peaks\) and the 1km local model \(right, detailed peaks\).](https://ss.rapidrecap.app/screens/F_PTJG6K2-E/00-13-34.png) ![Screenshot at 15:29: Final slide showing logos of various partners supporting the TEDxTrento event.](https://ss.rapidrecap.app/screens/F_PTJG6K2-E/00-15-29.png)