AI Revolutionizes Nowcasting, Paving the Way for Faster and Smarter Weather Forecasts

Artificial intelligence (AI) is transforming the way meteorologists predict the weather — particularly in the short term. Experts say that AI-powered nowcasting, which provides forecasts minutes to hours ahead, has the potential to drastically improve the accuracy and speed of weather warnings, thereby reducing casualties and economic losses caused by sudden, high-impact weather events such as thunderstorms, flash floods, and severe rainfall.

The Growing Need for Better Nowcasting

Many localized extreme weather events develop suddenly, intensify rapidly, and remain difficult to predict using conventional forecasting models. These phenomena — though limited in scope — often unleash devastating effects, from flash floods to landslides, that endanger lives and infrastructure.

Nowcasting focuses on the immediate timeframe, typically from the next few minutes up to six hours ahead. It combines real-time data from weather radars, satellites, ground-based sensors, and atmospheric observations to generate highly localized forecasts. This precision enables authorities to issue early warnings and mobilize rapid responses, a critical capability in the era of climate change and intensifying extreme weather.

The Role of Artificial Intelligence

For decades, meteorological forecasting has relied primarily on numerical weather prediction (NWP) — computer models based on the physical laws of atmospheric dynamics. However, these models are computationally intensive and often struggle with the fast-evolving, localized nature of short-term weather systems.

AI and machine learning are now revolutionizing the field by complementing physics-based models with data-driven algorithms that can learn complex weather patterns from vast datasets. These systems can detect and predict evolving storm structures, precipitation intensity, and wind shifts far faster than traditional models — in some cases, within seconds.

Recent research has shown that AI-enhanced nowcasting can outperform conventional systems in predicting severe convective storms, heavy rainfall, and cloud formation, leading to earlier and more precise warnings.

The WMO’s AI for Nowcasting Pilot Project (AINPP)

To accelerate the operational use of AI in forecasting, the World Meteorological Organization (WMO) has launched the AI for Nowcasting Pilot Project (AINPP) under the WMO Integrated Processing and Prediction System (WIPPS). The initiative, implemented in partnership with the World Weather Research Programme (WWRP), aims to bridge the gap between research innovation and real-world application, especially for developing nations.

As part of the project, the WMO and the Korea Meteorological Administration (KMA) jointly hosted the WMO AINPP Workshop from 24–26 September 2025 in Jeju, Republic of Korea, drawing more than 70 experts from National Meteorological and Hydrological Services (NMHSs), research institutions, universities, and major technology companies such as Google, Microsoft, and NVIDIA.

Collaboration Between Science, Policy, and Technology

“The rapid evolution of AI-driven nowcasting marks a pivotal step toward building more resilient societies in the face of extreme weather,” said Dr. David John Gagne, co-chair of the AINPP Steering Group and head of the Machine Integration and Learning for Earth Systems group at the National Center for Atmospheric Research (NCAR).

“By fostering collaboration between research and operations, ensuring equitable technology transfer, and engaging both public and private sectors, the global meteorological community is laying the foundation for faster, more accurate forecasts that can save lives and reduce disaster risk worldwide,” he added.

The workshop also emphasized the importance of inclusivity and capacity building — ensuring that AI innovations benefit developing countries that often lack access to advanced computing and data infrastructure. Participants discussed strategies for technology transfer, open-source sharing of AI models, and the creation of guidelines for operational use.

Bridging Research and Operations

A central theme of the discussions was how to move AI-based nowcasting tools from laboratories into real-world forecasting centers. The participants stressed the need for standardized evaluation metrics, containerized deployment systems such as Docker, and enhanced regional nowcasting centers capable of supporting local meteorological agencies.

Other recommendations included:

• Developing technical guidelines for integrating AI into operational systems;

• Promoting open-source collaboration to make advanced tools widely accessible;

• Training meteorologists and data scientists in AI methodologies;

• Establishing cross-sector partnerships with private tech companies and academia.

Emerging Trends in AI-Driven Weather Prediction

AI models used for nowcasting have advanced significantly in recent years. The latest research is moving from Convolutional Long Short-Term Memory (ConvLSTM) architectures to Transformer and Diffusion models, which are better at handling large-scale, multi-dimensional data.

Other key trends include:

• Integration of multi-source data, combining radar, satellite, and ground observations for richer inputs;

• Use of probabilistic ensembles to quantify forecast uncertainty;

• Linkages across nowcasting to medium-range forecasting, creating continuity across timescales;

• Optimization for real-time operational performance;

• Enhanced verification frameworks for model reliability and fairness.

One successful example cited during the workshop was KMA’s “NowAlpha-Diff” model, which uses diffusion-based AI to extend reliable motion prediction up to six hours, effectively overcoming the west-to-northeast bias that commonly affects mid-latitude forecasts.

The Path Ahead: From Innovation to Implementation

The workshop concluded that AI is no longer a futuristic add-on to meteorology — it is becoming an essential pillar of global early warning systems. However, experts cautioned that AI must be developed responsibly, with transparency and human oversight, ensuring that automated systems complement rather than replace professional forecasters.

The participants called for sustained international cooperation, particularly through the WMO framework, to ensure equitable access to AI-driven forecasting technologies. Strengthening partnerships among national services, private technology firms, and research institutions will be key to ensuring that AI nowcasting serves as a global public good.

As extreme weather events become more frequent and intense due to climate change, the integration of AI into forecasting systems offers a lifeline. The fusion of cutting-edge computation, meteorological science, and global collaboration promises to usher in a new era where life-saving information reaches communities faster than ever before.