AI and Sustainability: Transforming Climate Action

AI is reshaping weather forecasting, shifting it from a brute-force computational task into one of pattern recognition.

Conventional numerical weather prediction platforms rely on supercomputers to solve intricate atmospheric equations across global grids, processing terabytes of information yet still failing to capture the localised conditions that matter for business continuity.

Machine learning models take an alternative path. Rather than crunching physics equations in real time, AI systems draw on decades of historical weather records to uncover atmospheric patterns tied to specific outcomes.

This approach enables forecasts at 200-metre resolution while consuming only a fraction of the computing power demanded by traditional methods.

The transition is significant as climate change fuels more frequent and severe weather events that outpace the capabilities of conventional forecasting. Legacy models operate on kilometre-scale grids and refresh every six to twelve hours.

By contrast, AI-based systems deliver street-level insights updated in near real time, giving enterprises the ability to anticipate and adapt to weather risks with a level of precision commercial users could not previously access.

“AI-powered weather forecasting has the potential to revolutionise high-quality, high-resolution weather and disaster management solutions, particularly in this accelerating phase of climate change,” says H.E. Dr. Abdulla Al Mandous, Director General of the National Center of Meteorology UAE.

“By enhancing prediction accuracy and enabling hyper-local, real-time forecasts, this technology empowers better decision-making and strengthens resilience against climate challenges.

“We are excited to see how these innovations will shape the future of sustainability and drive more effective, data-driven solutions. The National Center of Meteorology UAE is rapidly embracing AI technologies for advanced weather forecasting and disaster management, positioning itself at the forefront of global efforts in weather and climate resilience.”

Weather-related disruptions account for 30% of flight delays according to Federal Aviation Administration data, costing airlines billions annually. 

Conventional forecasts can describe overall airport conditions but fall short when predicting runway-specific factors such as wind shear or visibility.

Ground transport encounters the same limitations, with localised fog or precipitation creating safety risks that broader regional forecasts fail to capture.

Behind G42 and Nvidia’s AI and Earth-2 partnership

G42 and Nvidia have built an AI-driven weather forecasting platform through the Earth-2 Climate Tech Lab in Abu Dhabi, delivering 200-metre resolution predictions powered by a generative downscaling model that captures small-scale regional physics to anticipate extreme weather events.

Inception, a G42 subsidiary collaborating with Space42, has adapted Nvidia’s CorrDiff system from the Earth-2 platform to provide highly detailed forecasts for urban environments.

Core42, G42’s digital infrastructure arm, underpins the deployment with Nvidia hardware.

The platform has already demonstrated its effectiveness with an end-to-end fog simulation across the UAE, addressing localised weather conditions that significantly impact industries from aviation to logistics.

Andrew Jackson, Chief Executive Officer of Inception, says: “For AI to be truly transformative, it should be an accessible tool for governments and industries worldwide.

“Through our collaboration with Nvidia, we are bringing cutting-edge forecasting capabilities within reach of those who need them most. Because CorrDiff is designed to adapt to local weather behaviours, this technology is not only improving forecasting for the UAE but can also be tailored for regions worldwide facing climate volatility.”

I forecasting systems adapt to local weather patterns across geographical regions.

Traditional models use global atmospheric equations that may not capture regional variations. AI systems learn from local historical data to improve predictions for specific areas over time.

Energy companies are turning to hyper-local weather forecasting to optimise renewable energy operations. 

According to the International Energy Agency, enhanced forecasting could boost wind energy efficiency by 15% through improved prediction of wind patterns.

Traditional renewable energy forecasting functions at kilometre-scale resolution, limiting its accuracy for individual sites.

In contrast, AI-driven systems deliver forecasts tailored to specific turbines or solar arrays, enabling more precise and efficient energy management.

How Nvidia physics AI models process climate data

Physics-based AI models interpret climate data by identifying atmospheric patterns through advanced data analysis instead of depending exclusively on complex mathematical equations.

This method accelerates forecast generation and lowers the demand for computational infrastructure compared with conventional forecasting systems.

Dion Harris, Senior Director of HPC and AI Factory Solutions at Nvidia, says: “Hyper-local weather forecasting is becoming crucial to determining patterns and people positioning as climates around the world undergo change. G42 is innovating using the Nvidia Earth-2 platform to drive access to actionable information in areas that need it most.”

The Nvidia Earth-2 platform delivers advanced computational frameworks for high-resolution climate modelling, enabling scalable processing of environmental data from satellites, ground sensors and atmospheric monitoring networks.

While traditional forecasting has depended on supercomputing resources accessible only to major meteorological agencies, AI-driven forecasting on Earth-2 makes these capabilities broadly available, effectively democratising access to next-generation climate insights.

Dion continues: “Weather forecasting has always required significant computational power, but AI is redefining what’s possible. With Physics AI models for high-resolution climate modeling, we can now generate faster, more detailed and more adaptive predictions. The scalability of accelerated computing allows us to process vast amounts of data efficiently, helping industries and governments better understand and respond to extreme weather patterns.”

Insurance companies leverage AI-powered weather forecasting to enhance operational efficiency and risk assessment.

According to Munich Re, weather-related claims account for 60% of global property insurance losses.

AI-driven forecasts enable insurers to evaluate risks at precise locations, moving beyond broad regional data to deliver tailored insights that improve underwriting and claims management.

This shift helps insurers better predict and mitigate weather-related damages, supporting more accurate pricing and faster response to claims.

Technology expansion targets climate vulnerable regions

G42 is pursuing the expansion of its hyper-local forecasting systems to climate-vulnerable regions in Africa, South Asia and Southeast Asia.

These areas face outsised impacts from climate change despite contributing relatively low emissions, making precise, localised weather forecasting critical for improving resilience and mitigating risks in vulnerable communities.

Agricultural applications offer significant potential in developing regions, where climate-sensitive farming practices depend heavily on accurate weather forecasts.

The Food and Agriculture Organization estimates that weather-related events cause approximately US$5 billion in annual crop losses globally.

Farmers reliant on rain-fed agriculture particularly need precise precipitation forecasts to optimize planting and harvesting schedules, helping to reduce crop failures and improve food security.

Andrew adds: “The ability to generate hyper-local predictions gives decision-makers the confidence to act faster, plan better and build more resilient systems. Beyond its impact on weather forecasting, this breakthrough has far-reaching applications in aviation, urban mobility, energy grid optimisation and environmental planning. Hyper-local insights can reduce flight delays, improve road safety, optimise renewable energy distribution and support climate-adaptive urban development.”

Accurate precipitation forecasts are crucial for water resource management, allowing utilities to optimise reservoir operations and maintain water supply reliability.

Early warning systems powered by precise weather predictions also play a vital role in protecting vulnerable populations by providing advance alerts for extreme weather events, ultimately helping to save lives in climate-impacted regions.

Dion concludes: “The scalability of accelerated computing allows us to process vast amounts of data efficiently, helping industries and governments better understand and respond to extreme weather patterns.”