The use of artificial intelligence can significantly improve rainfall prediction and help meteorologists in studying weather patterns more effectively in the future.
At the ongoing weather summit in the United Arab Emirates, experts are exploring the implementation of AI-driven technology in weather forecasting and cloud seeding.
The event, AI for Weather Prediction: Advances, Challenges & Future Outlook conference, is being held in Abu Dhabi from Sept. 9-11.
On the sidelines of the event, Monica Youngman from the US National Weather Service commented that choosing the right AI model is central to this process.
"It depends on the question you're trying to answer. AI models exist at global, regional, and local scales, and you have to carefully target datasets to train models for specific use cases. Foundational models can be fine-tuned, but they must be used correctly," she was quoted as saying by the Khaleej Times.
She noted that while AI can be fine-tuned, it must be used with care. Youngman also shared her views on forecast accuracy. “The most useful state is the three-to-seven-day forecast range. AI models struggle further out because they aren't based in physics,” she said.
According to Youngman, for very short-term forecasts, or nowcasting, AI is improving. Still, traditional models remain critical for a broader understanding of weather patterns.
One of the sessions at the summit focused on advanced Earth system prediction using high-quality data from weather stations, satellites, IoT devices and maps, according to a Khaleej Times report. These tools help experts study weather patterns more closely.
“The life span of clouds is very short, so if you have multiple clouds, AI can help forecasters choose the optimum locations by identifying certain parameters. Now, operations are manual. By using AI, the operator can look at certain aspects of the clouds and decide, as AI guides them, where to go,” Omar Al Yazeedi, deputy director general of the National Centre of Meteorology, was quoted as saying by Khaleej Times.
The summit outlined that AI is meant to support weather prediction and modification. Ian Lisk, Chair of WMO SERCOM, noted that current models have limits and urged caution for the application of AI.
“For some weather parameters, like thunderstorms or convective storms, even traditional physics-based models struggle,” he said. “AI models have similar resolution issues. One key challenge is the training data, especially in a changing climate.”
The officials at the summit also noted that this AI-powered adoption of a model for improved forecasting should be inclusive in nature.
“Many countries lack the capacity to use AI, making it WMO's duty to ensure no one is left behind,” Ko Barrett, deputy secretary general of WMO, noted.
Essential Business Intelligence, Continuous LIVE TV, Sharp Market Insights, Practical Personal Finance Advice and Latest Stories — On NDTV Profit.
