Forecasters close in on north Atlantic hurricanes

Expect a fiercer-than-normal hurricane season this year, US agencies warned last month – but these seasonal forecasts are notoriously fickle. Now, a computer model has come impressively close to matching real-world annual hurricane counts in the north Atlantic. It could improve seasonal forecasts, used by emergency managers and the insurance industry to assess risks and by farmers to predict rainfall.

Weather forecasts and climate simulations are made with similar computer models operating on different timescales. Weather forecasters model regional changes for the coming hours to days; climate simulations are global and span years. Seasonal hurricane forecasts fall between the two, making them difficult to model. Current seasonal hurricane forecasts from the US National Hurricane Center and Colorado State University in Fort Collins are based on statistics rather than computer models.

Jan-Huey Chen and Shian-Jiann Lin of the US National Oceanic and Atmospheric Administration's Geophysical Fluid Dynamics Laboratory in Princeton, New Jersey, created a new global model, which divides the atmosphere into cells, 625 square kilometres each. These are smaller than those in standard climate models but larger than those in weather models, making the new model better able to replicate climate effects on regional processes such as hurricane formation.

The acid test of computer models is how closely they replicate conditions of past years. To test theirs, Chen and Lin fed it data of global sea-surface temperatures gathered between 2000 to 2010; sea-surface temperatures are critical in hurricane formation. Then they counted the number of times that surface winds produced by their model reached hurricane force (119 kilometres per hour).

Their model averaged 1.43 fewer hurricanes each year than were recorded in the north Atlantic over those years, but closely tracked year-to-year variations in hurricane counts. Crucially, the model reproduced the sharp drop in hurricanes from the record year of 2005 to the unexpectedly quiet 2006, and the jump from 2009 to 2010.

"This forecast is obviously very skilful since 2000," says Phil Klotzbach, who works on the Colorado State forecasts. But he cautions that it might not work as well for earlier years, when hurricane numbers were less sensitive to sea-surface temperatures, possibly because of shifts in the north Atlantic oscillation in the mid-1990s. Chen plans another modelling run starting in 1985 to test a broader range of conditions.

Climatic interconnectivity — Eurasian autumn snow coverage affects ...

Results showed that autumn snow cover from northern Scandinavia to the West Siberian Plain is most significantly associated with winter temperatures over the interior of North America. More (less) frequent snow cover over this region is related to lower (higher) winter temperatures over the interior of North America in January, extending to the eastern and southern United States in February.

The North Atlantic oscillation, climatic significance and environmental impact

The North Atlantic oscillation, variability and interactions with the North Atlantic Ocean and Arctic Sea ice

The North Atlantic oscillation, climatic significance and environmental impact

The complete guide to climate change

Hydrological, Socioeconomic and Ecological Impacts of the North Atlantic Oscillation in the Mediterranean Region