Harnessing the power of AI to assess the impact of climate change

As climate change continues to intensify, the world is facing increasing numbers and severity of climate- and weather-related disasters. UN Member States need to develop urgent mitigation and adaptation measures based on reliable climate change projections.

However, these projections can be difficult to obtain. Climate change impact assessment models are the primary means of estimating future climate-related impacts, but they are complex computational models that require expertise to run and interpret.

Such expertise and computational requirements are rare in many regions and communities. The lack of viable impact assessments leaves communities at a disadvantage and more vulnerable to climate change.

A new report by UN scientists urges the global climate action community and UN member states to consider artificial intelligence (AI) as a promising solution to the problem: AI is flexible, computationally efficient, and can run on web- and cloud-based services.

As AI continues to grow in popularity across a range of applications, it is likely that a larger audience will be able to interact with and interpret AI output than traditional climate change impact assessment models.

However, AI has been underutilized in the field of climate change impact assessments, which is why researchers from the United Nations University Institute for Water, Environment, and Health (UNU-INWEH) decided to explore its capabilities to conduct rapid, flexible and reliable climate change impact assessments in 46 cities across the United States.

Their findings demonstrate the power of this approach as an alternative to traditional, computationally intensive and time-consuming methods for assessing local-level climate change impacts.

“AI is a powerful tool used in almost every sector, from imaging to medicine, so there is no reason why it couldn’t also be used to build societal resilience to climate change,” said Dr Renee Obringer, Research Fellow in Urban and Interdependent Infrastructure Systems at UNU-INWEH, who led the study.

The report projects that across the 46 cities modeled, water demand could increase by up to 15% and electricity demand by up to 20%. Midwestern US cities could see an average increase in electricity use of 20% if global temperatures rise by 2.0°C above pre-industrial levels. For Chicago, this equates to 745,000 MWh/month, equivalent to the amount of electricity used annually by more than 26,000 UK residents, 98,000 Indian residents, or 292,000 Nigerian residents.

In terms of water consumption, the study found that Midwestern US cities could experience a 7.5% increase above pre-industrial levels at 2.0°C. The average US household uses more than 1,100 litres of water per day, so increasing this by 7.5% would amount to an additional 85 litres per day per household and 57 million litres per day across cities. This is the equivalent of roughly 23 Olympic-sized swimming pools consumed in one city in a single day.

California cities are expected to see large increases in electricity demand, but relatively small increases in water demand. However, some cities in the southern U.S. may see reduced water demand in summer due to increased precipitation and humidity from a warming climate.

In their report, UNU scientists show how they can make low-cost, easy-to-understand predictions by using the concept of “climate analogues,” which matched each US city studied with another city in North America whose current climate matches the best estimate of the US city’s future climate.

For example, their findings show that New York City’s future climate is most similar to the current climate of Jonesboro, Arkansas, a city in the southern U.S. By leveraging these similarities, the researchers were able to eliminate the need for complex climate models, further improving the accessibility of their AI approach.

The use of these analogues is a novel approach that streamlines the process and is particularly useful for regional climate change impact studies. “Analogues reduce the computational cost of the AI ​​impact assessment process and eliminate the need for users to obtain large-scale, complex future climate projections. This is an ideal way to minimise cost and effort for countries, regions and cities that lack climate simulation expertise and capacity,” said Dr. Obringer.

The authors make several recommendations for expanding the use of AI in climate change impact assessments, including the development of web- and cloud-based services that could be particularly useful in the Global South. But the report cautions that AI relies on high-quality, readily available data, which is a problem in many parts of the world.

Therefore, while society aims to build resilience to climate change through equitable mitigation and adaptation policies, it is important that relevant data and models remain open and freely accessible so that local communities can understand the impacts of climate change in their areas.

“This report is just one example of how the world can benefit from the technological advances of the Fourth Industrial Revolution and address challenges of equity and sustainability,” said Professor Kaveh Madani, Director of UNU-INWEH.

“While AI has its own risks and limitations, it also has significant benefits that should not be ignored, especially when it comes to addressing the digital divide and inequality. Lack of computing infrastructure should not prevent countries from taking appropriate climate action. AI can help the Global South and disadvantaged communities overcome significant barriers and improve their planning to deal with climate change and extreme weather.”