Weather & Climate Forecast Conference (WCFC)2025
Pioneering a New Era of Weather & Climate Forecasting Through AI and Global Collaboration
On December 16, 2025, the inaugural Weather & Climate Forecast Conference (WCFC) 2025, organized by the WNI Wx Bunka Foundation, was held at the Tokyo International Forum. Under the theme "New Value for the Weather Industry, Unlocked by AI," the conference brought together leaders from academia, industry, and national meteorological and hydrological services to discuss the modernization of weather infrastructure and advances in forecasting technology across the Asia-Pacific region. The ultimate goal of this conference is to cultivate a new "weather culture" — one that prioritizes disaster risk reduction and societal resilience through practical, action-oriented initiatives.
WNI Wx Bunka Foundation — Representative Director, Jiro Miyabe
1. A Paradigm Shift: AI-Powered Weather Forecasting
The conference spotlighted a transformational shift underway across the industry. Artificial intelligence (machine learning) has now surpassed the long-standing "gold standard" of physics-based numerical weather prediction (NWP) models.
- Dramatic Gains in Speed and Accuracy: Google DeepMind's latest model, the "Functional Generative Network (FGN)," has reduced computation time that once took hours on a supercomputer to just minutes on a single GPU. Furthermore, AI has achieved state-of-the-art accuracy in predicting tropical cyclone intensity and structure — a domain where conventional models have historically struggled.
- The Power of 1,000-Member Ensembles: Leveraging AI's low computational cost, researchers have successfully implemented 1,000-member ensemble forecasts. This enables forecasters to detect extreme weather events at probabilities as low as 1%, providing early warning signals for rare but catastrophic disasters that were previously difficult to assess.
Google DeepMind — Research Scientist, Ferran Alet
- A Breakthrough in Typhoon Forecasting — A New Era of Precision: Weathernews, Inc. (WNI)'s Global Storm Center demonstrated that Google DeepMind's FNV3 model, during the 2025 season, reduced track errors by 40% and intensity errors by 20% compared to conventional benchmarks. Since August 2025, WNI's official typhoon track forecasts — powered by ECMWF's AIFS model — have consistently maintained approximately 20% higher accuracy than those issued by major international agencies such as JTWC and JMA. AI models have also shown remarkable stability from the earliest stages of typhoon development, providing the reliable information forecasters need to make swift and well-informed decisions.
"The Potential of AI as Seen Through Typhoon Track Forecasting"
Weathernews, Inc. (WNI) Global Storm Center — Yoshihiro Konno
2. Perspectives from National Meteorological and Hydrological Services (NMHS)
In the panel discussion, representatives from the national meteorological and hydrological services (NMHS) of Indonesia (BMKG), the Philippines (PAGASA), Thailand (TMD), and Vietnam (VNMHA) shared the significant challenges and high expectations surrounding AI integration in tropical regions.
- Responding to Intensifying Disasters: Countries across Southeast and South Asia sit on the front lines of climate change, facing increasingly frequent and severe typhoons, flash floods, and localized heavy rainfall. Strengthening early warning systems (EWS) through "impact-based forecasting" is the top priority for protecting lives and property.
- Addressing the Limitations of Numerical Forecasting: Conventional physics-based NWP models often struggle with complex tropical convection and 0–3 hour "nowcasting." Stakeholders position AI not as a replacement for physics-based models, but as a "transformative layer" that improves forecast consistency and bridges the gap between meteorological data and actionable guidance.
- The Need for Global Collaboration: Deploying these AI solutions requires substantial investment in GPU-based high-performance computing and capacity building in specialized AI engineering skills. Strong collaboration with major technology companies and international partners is urgently needed to provide the necessary cloud scalability and "Explainable AI (XAI)" frameworks that ensure the reliability of alerts.
"Panel Discussion with Representatives from National Meteorological Services"
3. Real-World Implementation: AI as a Decision-Support Agent
The conference also showcased how AI is evolving from a mere "forecasting tool" into an "intelligent agent" that guides human decision-making.
- AI Weather Agent: WNI unveiled a "Weather (Otenki) Agent" powered by large language models (LLMs). By integrating 20 years of "human-sensed" data drawn from over 100 million user reports with high-resolution analysis data, the AI can respond to questions such as "Should I bring an umbrella?" or "How will this affect our supply chain?" — in natural language.
- Creating Business Value: By standardizing "weather intelligence" as a decision-making signal, weather data can be systematically integrated into inventory management, demand forecasting, risk management, and operational planning. Through this integration, weather information transforms from a simple forecast into a direct driver of economic value and social benefit — including disaster mitigation.
Weathernews, Inc. (WNI) — AI CoE, Akihiro Morimoto
4. A Vision for the Future: Co-Creation and the Role of Observations
The University of Tokyo (ClimCORE) emphasized that "observations are the only tool connecting the real atmosphere to the digital world," calling for the continued importance of maintaining high-quality observational networks and "reanalysis" data as the foundation for training robust AI models.
The conference concluded with a vision for "division of roles" in the age of AI:
- Big Tech Companies and International Organizations: Develop large, high-precision global AI models.
- National Meteorological and Hydrological Services (NMHS) and the Private Sector: Customize regional and application-specific AI models tailored to local terrain and specific societal needs.
University of Tokyo Institute for Advanced Study — ClimCORE Senior Program Advisor, Kenichi Kuma
A Call to Participate
WCFC 2025 underscored a strong conviction that by aligning emerging AI technologies with the needs of national meteorological and hydrological services (NMHS) and society at large, it is possible to strengthen early warning systems and save lives.
The next conference will address the key challenges identified through these discussions — including computational resource constraints, the need for high-resolution, region-specific, and application-specific models, capacity building for effective AI adoption, and the integration of AI technologies suited to each country's meteorological hazards. Solution-oriented dialogues among Asian meteorological authorities, AI innovators, academic institutions, and private-sector companies such as Google DeepMind and Weathernews, Inc. are anticipated, with the shared goal of advancing disaster preparedness and response across the international community.
Closing Remarks — Weathernews, Inc., Executive Officer, Daisuke Abe
"Keynote Address" |
"Spatio-temporal Characteristics of Summer Rainfall in Upper Northeastern Thailand under ENSO Influence: Evidence from Radar Data and Numerical Modeling" |
"Toward Applying AI for Rainfall Estimation and Prediction in Vietnam" |
"The Role of the Academe in Meteorological Research in the Philippines" |
"Challenges and Opportunities of Climate Adaptation in Thailand" |
"Applications and Challenges of AI Weather Models in the Asian Region" |
"Applications and Challenges of AI Weather Models in the Asian Region" |