This is how AI predicts the next extreme weather events

Learn how AI models surpass traditional methods in predicting climate effects with greater precision and detail. The technology that will change how we deal with the weather.
Wind energy on Global Wind Day

What if we’d been able to predict the unusual rains that flooded the Arabian Peninsula or China last spring? How would our daily lives change if we could anticipate extreme weather events?

In a world increasingly affected by climate change, anticipating these unusual weather events is crucial. Traditional weather forecasting tools, although effective to a certain extent, don’t always manage to accurately predict the complexities and magnitude of phenomena such as hurricanes, heatwaves, or torrential storms. This is where Artificial Intelligence (AI) comes into play, revolutionising our ability to understand and anticipate climate behaviour with unprecedented precision.

What will I read about in this article?

  • AI algorithms and climate models
  • Improvements achieved by AI
  • Success stories

 

How do artificial intelligence algorithms predict the effects of climate change?

An accurate weather forecast not only helps us decide what to wear when leaving the house in the morning but is also vital for warning us about possible extreme weather events. That information provides communities with crucial time to prepare adequately.

AI related to climate change uses advanced algorithms that can analyse vast amounts of historical and current atmospheric data. These technologies are capable of identifying patterns and correlations that might go unnoticed by the human eye.

One of the greatest advances it offers is the ability to make real-time predictions. Through the use of neural networks and other deep learning models, AI systems can be continuously updated with new information, adjusting their predictions almost instantly. This is invaluable for governments and emergency agencies, enabling them to make more informed and rapid decisions, potentially saving thousands of lives and reducing material damage.

An illustrative case of the effectiveness of AI related to climate change is Hurricane Harvey in 2017. While traditional models underestimated the amount of rainfall this devastating hurricane would bring, an AI-based model developed by IBM and its project "The Weather Company" was able to predict the historic rains more accurately. This information allowed local and federal authorities to better coordinate their response and mitigation efforts, demonstrating the transformative power of these new technologies.

 

How does AI improve traditional climate prediction models?

AI algorithms, especially those based on machine learning and neural networks, are helping to overcome these limitations in several ways:

  • Much more detailed and accurate climate models: They can analyse and process massive amounts of data in real time. This includes historical data, satellite images, atmospheric measurements, and even information from globally distributed sensors.
  • Detecting correlations and trends: AI techniques like deep neural networks are extremely good at identifying complex patterns in large datasets.
  • Continuous learning: As more data is collected, AI models can learn and adapt, continually improving their predictions. This continuous learning allows AI-based climate systems to become more accurate as they are fed new information.
  • Creating high-resolution climate simulations: These are much more detailed than those possible with traditional methods. This is crucial for accurately predicting local phenomena such as severe storms or urban microclimates.
  • Multidimensional data integration: AI can combine data from various sources such as satellites, weather stations, and ocean buoys to create a holistic view of the climate system.

Imagine a hurricane forming in the Atlantic. A traditional climate model might predict its general trajectory but with certain inaccuracies regarding its intensity and exact point of impact.

In contrast, an AI-powered model can analyse real-time data from aircraft, satellites, and ocean buoys, identifying patterns in wind speed, water temperature, and atmospheric pressure. This detailed analysis allows for more accurate predictions of the hurricane's trajectory, strength, and potential impact, giving authorities more time and better information to act.

 

Success stories of artificial intelligence in relation to climate change

Chile and drought monitoring

Drought is a constant threat to agricultural producers in Chile. To mitigate this risk, the Agricultural Research Institute (INIA) of the Chilean Ministry of Agriculture has developed an innovative AI technology related to climate change that allows agricultural droughts to be predicted three months in advance.

This tool uses mathematical algorithms and satellite images to offer forecasts with 95% accuracy, helping authorities make informed decisions and reduce the negative impact on agricultural production. During the presentation of this service, Undersecretary of Agriculture Ignacia Fernández emphasised the importance of these predictive tools for ensuring continuity of production and supporting food security.

 

USA and wildfire prediction

California has long been plagued by devastating wildfires, such as the tragic Camp Fire in 2018, which remains the deadliest and most destructive in the state's history. Although fewer fires were recorded in 2023 compared to previous periods, constant vigilance remains crucial.

Cal Fire, the state's fire department, has teamed up with the University of San Diego's AlertCalifornia system to develop a pioneering project that uses AI to alert firefighters to nascent fires. This technology scans images from strategically distributed cameras throughout the state, looking for signs such as smoke. When something unusual is detected, the system immediately alerts local officials who review the images and determine if intervention is necessary.

This summer, the system was tested at six Cal Fire emergency command centres across California. According to the Los Angeles Times, said system correctly identified 77 fires before authorities received calls about them, allowing firefighters to respond quickly and prevent the fires from spreading out of control.

 

Canada and hurricane preparedness

In 2023, an AI innovation from Google DeepMind marked a milestone in weather prediction by accurately forecasting the entry point of Hurricane Lee in Canada. The GraphCast tool, designed specifically to improve predictions of extreme weather events, managed to predict the hurricane's trajectory nine days ahead of traditional methods, allowing authorities to take necessary preventive measures.

This advanced AI technology uses machine learning and graph neural networks to make high-precision forecasts. Specifically, it analyses variables from the Earth's surface (temperature, wind speed and direction, and sea level pressure) and the atmosphere (humidity, wind speed and direction, and temperature).

Although GraphCast's reports are not as detailed as those of traditional weather prediction systems, it has been shown to be more efficient in predicting extreme temperatures, tracking hurricanes, or detecting potential floods.

 

Southern Europe and heatwave anticipation

The summer of 2023 in southern Europe was marked by a record number of days of extreme heat, with temperatures ranging from 38°C to 46°C, even exceeding 46°C on several occasions.

The difficulty in predicting heatwaves means that authorities struggle to adopt adequate preventive measures

Thus, the ISSUL project, funded by the Marie Skłodowska-Curie Actions, focused on developing advanced prediction models that extend weather forecasts beyond two weeks, but without covering an entire season

The ISSUL project utilised machine learning to improve the prediction of heatwaves in these countries. With optimisation and regression algorithms, they managed to accurately predict the monthly average temperatures of this territory one month in advance.

However, predicting the regional intensity of heatwaves remains a challenge, as the model didn’t achieve accuracy in this aspect. Despite this, researchers identified recurring sets of optimal predictors, suggesting opportunities to improve the predictability of extreme temperatures in the future.

 

 

Benefits of early prediction

The ability to predict extreme weather events in advance brings numerous benefits. To begin with, it allows communities and governments to better prepare, minimising loss of life and reducing damage to infrastructure.

In the case of renewable energy companies, for instance, they can optimise the operation of wind and solar farms by anticipating adverse weather conditions. There are also water treatment solutions that can be adjusted based on drought or flood forecasts, ensuring a constant and safe supply. Or the case of resilient infrastructure, which, designed with the help of AI predictive models, can better withstand the impact of extreme events, protecting both people and investments.

In summary, artificial intelligence related to climate change is not only redefining our ability to predict extreme weather events, but also providing essential tools to build a safer and more sustainable future. As we advance in this digital age, the integration of AI into climate models promises to be one of the most powerful weapons in our fight against climate change and its devastating effects.