AI Predicts Floods, Yet Cannot Eliminate Disaster Risk for Areas

Photo : Reuters

London: In September, severe floods swept through parts of Europe, leaving devastation in their wake. The scale of the destruction surprised many, even though advanced forecasting systems had predicted the heavy rainfall. These systems, bolstered by artificial intelligence (AI), accurately forecasted the rains, yet the resultant impacts in affected areas were unforeseen, underscoring the complexities of managing increasingly frequent extreme weather events. AI technology has transformed weather forecasting by leveraging statistical methods to analyze extensive historical data. This approach has proven to be more cost-effective compared to traditional numerical weather prediction models. It enables meteorologists to deliver more specific predictions, especially for events such as urban flooding or in challenging terrains like mountainous regions.

One notable example of AI in weather forecasting is GraphCast, a machine learning model funded by Google. This model, which utilizes reanalysis data—essentially a compilation of historical weather forecasts refined with modern techniques—has been shown to outperform conventional forecasting models. However, experts warn that the effectiveness of AI in forecasting is contingent on the quality of input data. In instances where there is a scarcity of data, or when extreme weather events become more common or occur at unpredictable times, predicting disasters becomes increasingly challenging. Andrew Charlton-Perez, a professor of meteorology at the University of Reading in the UK, stated, "In some cases and for some variables, AI models can outperform physics-based models, but the reverse can also be true." He highlighted the necessity of integrating AI forecasts into existing forecasting frameworks to enhance their reliability and precision.

Stay informed with the latest news. Follow DXB News Network on WhatsApp Channel 

Since the beginning of the year, the European Centre for Medium-Range Weather Forecasts (ECMWF) has been employing an AI-driven forecasting model called the Artificial Intelligence/Integrated Forecasting System (AIFS). This system produces rapid multiple predictions and long-term forecasts for weather events, such as cyclones and heatwaves. Ahead of the September floods, ECMWF's forecasts indicated potential rainfall of 300-400 millimeters (11.8-15.7 inches), which was subsequently realized. Despite these accurate predictions, experts emphasize the importance of effective communication in conveying the severity of potential disasters. Shruti Nath, a postdoctoral research assistant at Oxford University, noted, "Even if weather models capture extreme events, there's a reasonable degree of uncertainty due to their rarity, often classified as a one in 150- to 200-year event." Nath underscored the necessity of presenting warnings in a way that reflects the possible severity of impacts on people, thus enabling them to understand the potential costs of inaction versus taking preventive measures.

A recent report from the European Environment Agency (EEA) revealed that Europe is facing urgent climate risks that are surpassing current policies and adaptation efforts. The EEA warned that extreme weather events, including heatwaves, droughts, wildfires, and floods, will escalate even under optimistic global warming scenarios, significantly affecting living conditions across the continent. Following the recent floods, Janez Lenarcic, the European Commissioner for Crisis Management, remarked that these extreme weather events, once deemed rare, are now occurring almost annually. He described the situation as a stark reflection of the global reality of climate breakdown infiltrating the daily lives of Europeans.

Some tech entrepreneurs are concerned that Europe is not adequately prepared for these challenges. Jonas Torland, co-founder of Norway-based 7Analytics, which specializes in flood and landslide prediction models, pointed out that risk managers in the U.S. are generally more experienced in assessing environmental hazards. In contrast, European authorities often lack the necessary readiness to effectively address these issues. Torland explained that substantial expenditures frequently occur without adequate data support for informed decision-making, and governments tend to rely on outdated data providers and consultants rather than investing in advanced AI solutions that could enhance their predictive capabilities.

The challenge extends beyond the models themselves. Implementing complex AI models requires significant computing power and frequent updates, which can be time-consuming and resource-intensive. For example, a fine-grained 1-by-1 meter grid used by 7Analytics for predictions is over 100 times more detailed than a broader 10-by-10 meter grid, necessitating substantially longer processing times. Moreover, the high computational demands associated with these AI models contribute to increased energy consumption and environmental impact, leading to greater emissions that further exacerbate the climate crisis. Some tech giants, including Microsoft and Google, are exploring nuclear power as a sustainable energy source for their extensive data centers.

Experts argue that, beyond enhancing forecasting technologies, there is a pressing need for investment in physical infrastructure solutions, such as designated areas for floodwater storage and improved early warning systems. Authorities must also prioritize limiting development in flood-prone regions and uphold commitments to reduce greenhouse gas emissions. Friederike Otto, a senior lecturer at Imperial College London, emphasized that the issue is not one of technology or knowledge, but rather a question of political will and implementation. She stated that as long as the world continues to burn fossil fuels, the root cause of climate change will persist, leading to increasingly severe weather events that threaten lives and livelihoods. To combat this trend, she advocated for a transition from oil, gas, and coal to renewable energy sources.

In summary, while AI presents a promising avenue for enhancing weather predictions, the ability to effectively mitigate the risks associated with extreme weather remains a complex challenge that requires a multifaceted approach, including investment in technology, infrastructure, and communication strategies.