Advances in artificial intelligence (AI) have the potential to significantly improve the accuracy of weather predictions, increasing profits for companies that rely on them.
Google introduced new AI model It can produce accurate, large-scale weather forecasts at a lower cost than traditional methods. Experts say AI models have the potential to streamline operations, reduce costs and enhance decision-making across industries by providing more accurate and timely weather information.
“AI-based weather forecasts are useful because they can perform predictions much faster than traditional physics-based weather forecasts and are generally cheaper, as Google's recent research shows.” marty sullivanPostdoctoral Fellow in Earth and Atmospheric Sciences. Cornell University, he told PYMNTS.
AI could be better at predicting
Google's “Scalable Ensemble Envelope Diffusion Sampler” (SEEDS) model works similarly to popular AI tools such as OpenAI's ChatGPT to improve weather forecasting by quickly and affordably creating multiple possible weather scenarios. It's improving. Researchers say that this method Science Advance Journalnote the improved efficiency compared to the traditional model.
Weather forecasting typically uses physically-based methods. Take different measurements and combine them Predict a variety of potential outcomes instead of one prediction.
AI can be useful when predictions cannot be fully explained by known or small numbers of variables. In this case, the physics of weather, Phil Siegelfounder of the AI Nonprofit Center for Advanced Preparedness and Threat Response Simulation (captor)he told PYMNTS.
“Often these models can use data in ways that are difficult to model and difficult to incorporate into purely computationally oriented models,” he added. “The flip side is that AI might think it’s recognizing patterns that don’t exist. This is where machine learning comes in: making sure results are fed back so the AI can improve. .”
New research supports the idea that AI can outperform traditional predictive methods. Researchers at the University of Reading recently We evaluated the performance of an AI-based weather model Traditional physics-based models were used to predict Storm Ciaran, a severe storm that hit northern and central Europe in November and caused power outages that killed 16 people and affected more than 1 million homes in France. It was done by comparing.
The study found that weather forecasting using machine learning can achieve similar accuracy to traditional methods, but faster, at lower cost, and with less computational complexity.The research results were published in a magazine climate and atmospheric science.
The study included four AI models developed by companies such as Google, Nvidia, and Huawei. These models accurately predicted the storm's intensity and path 48 hours into the future. The results were similar to previous models and effectively captured important atmospheric conditions, such as the position of the storm relative to the jet stream.
“AI is transforming weather forecasting before our eyes,” study leader Andrew Charlton Perez said in the paper. news release. “Two years ago, modern machine learning techniques were rarely applied to weather forecasting. We now have multiple models that can generate 10-day global forecasts in minutes.
“We can learn a lot about AI weather forecasting by stress testing it with extreme events like Storm Ciaran,” he added. “We can identify their strengths and weaknesses and guide the development of even better AI predictive technology to protect people and property. This is an exciting and important time for weather forecasting.”
Advantages of sunny weather
Even small improvements in weather forecasting capabilities can boost a company's bottom line. AI-powered weather data provides valuable signals that companies can use to strengthen supply chain resiliency and improve forecast accuracy of demand, material availability, and other critical logistics factors . gel tikaraExecutive Vice President of Engineering at a Supply Chain Management Company Kinaxishe told PYMNTS.
“Aided by geolocation and historical patterns, retailers can adjust inventory levels in real time, stocking up on summer essentials like sunscreen just before a heat wave, or stocking up on scarves and warm clothing ahead of a snowstorm. “By doubling orders, we can minimize the risk of empty store shelves,” he said, “while keeping customers happy.”
Similarly, businesses can use AI-driven predictions to reroute shipments to avoid anticipated delays and damages caused by severe weather such as snowstorms, ensure smoother operations, and reduce the impact of disruptions. It can be done, Tikala said.
“Weather is an unavoidable wildcard for both customers and businesses, but by leveraging a combination of cutting-edge predictions and ‘what ifs’, scenario planning capabilities can help supply chains one day rival Mother Nature.” “It could give us the edge we need,” he added.