IPMA - UWeather | Utilities, Data & AI

UWeather – IPMA (Portuguese Institute for Sea and Atmosphere)

Utilities, Data &AI

Python, TensorFlow, HuggingFace, Google BERT

Context and Goal‍

The IPMA sought to modernise the way descriptive weather forecasts were produced. Traditionally created manually by meteorologists, these forecasts demanded significant time and expert resources. The main goal was to develop an intelligent assistant capable of transforming complex meteorological data into vivid, engaging descriptions, supporting the work of professionals while enhancing the public’s experience.

Challenges

• Automating the generation of descriptive forecasts through Artificial Intelligence and Natural Language Processing (NLP).
• Overcoming the lack of similar solutions in Portuguese, ensuring not only local value but also potential for expansion across other Portuguese-speaking countries.
• Designing a scalable methodology that could be applied to other domains where weather conditions are critical, such as fishing, water sports, agriculture, and renewable energy (wind and tidal).

Solution‍

agap2IT developed UWeather, an application powered by advanced AI models (TensorFlow, HuggingFace and Google BERT) that interprets raw meteorological data and converts it into natural and easily understandable narratives. The solution was designed to assist meteorologists in their daily work, not to replace their expertise, but to extend their capacity for analysis and communication.

Results and Impact

• Increased efficiency in producing descriptive forecasts.
• Greater clarity and accessibility of weather information for the public.
• A pioneering model in the Portuguese language, with potential for international adoption across the Lusophone world.
• Opportunities for cross-sector application, enhancing the solution’s value beyond traditional meteorology.

‍

‍

‍