Using deep learning, analysts at IBM Research Ireland, Baylor University, and the University of Notre Dame built a system that far outpaces existing prediction models.
Costly and sluggish traditional platforms require a supercomputer to calculate how tides, winds, and the ocean’s varying depths influence the speed and height of waves. The new deep learning-enhanced framework, however, generates forecasts up to 12,000 percent faster than conventional designs, according to IBM Research member Fearghal O’Donncha, who also tipped “a vastly increased” set of data input.
“Accurate forecasts of ocean wave heights and directions are a valuable resource for many marine-based industries,” O’Donncha wrote in a blog post. “Many of these industries operate in harsh environments where power and computing facilities are limited. A solution to provide highly accurate wave condition forecasts at low computational cost is essential for improved decision making.”
Even artificial intelligence needs to learn, though. The team put the time-honored Simulating WAves Nearshore (SWAN) model to work—generating training data (four years of forecasts, from April 2013 to July 2017) for their deep learning network. A roaring success, the AI replicated images of more than 3,000 wave heights and periods with fewer errors than SWAN.
“Despite the huge reduction in computational expense, the new approach provides compar