Net Hurricane Activity, Atlantic Sea Surface Temperatures, Salinity Factors in Colorado State University’s Landfall Prediction Model

Hurricane records since 1900, estimates of damage potential and a theory that a period of more intense hurricane activity may be beginning all contribute to a new statistical model that predicts hurricane landfall probabilities along the U.S. coast.

Developed by William Gray, Colorado State University professor of atmospheric sciences, the model charts landfalls along the entire coastline from Brownsville, Texas to Eastport, Maine; 11 arbitrary segments of coast; and 100 kilometer (62 mile) stretches of coast, assigning each a probability.

These probabilities, expressed as percentages, indicate the chance of occurrence in 1998. For example, a 2 percent chance that a hurricane will make landfall upon a given 100-kilometer coastal section means there is a 1-in-50 chance that a hurricane will strike that area this year.

An important element in Gray’s model is the Atlantic Ocean "conveyor belt," or Atlantic thermohaline. This circulation system moves Atlantic waters northward from the vicinity of the Caribbean to an area east of Greenland. There, the current sinks to deep levels, moves southward and flows into the South Atlantic Ocean and beyond.

Warm water in the conveyor belt – perhaps one degree Fahrenheit above the average North Atlantic sea surface temperature of about 45 F in the regions west of the British Isles – is associated with more intense hurricanes and more landfalls along the eastern seaboard, Gray said. Cool water tends to dampen hurricane activity.

During the period from the late 1920s to the late 1960s, generally warmer sea surface temperatures led to processes that produced more intense hurricanes than did the quarter-century periods from 1900-25 and 1970-94, Gray’s data show.

Gray’s model requires a seasonal forecast of "Net Tropical Cyclone" (NTC) activity, a statistical averaging of overall hurricane activity for a given year, in combination with sea surface temperature anomalies in the North Atlantic. Gray calls the latter "a good proxy measure for the strength of the Atlantic Ocean thermohaline circulation."

An increasing NTC forecast number and strengthening thermohaline circulation increases the probability of major or intense (Category 3-5 on the 1-5 Saffir-Simpson scale) hurricanes’ landfall.

"When the season’s especially active you get many more intense storms making landfall than you would during an average season," he said. "In general, the more hurricane activity you have, the greater the probability of landfall.

"While it’s been suggested that our general forecasts (Gray heads a research team that has been issuing seasonal hurricane forecasts for the Atlantic Basin for the past 15 years, with above- average success) have no meaning in relation to landfall, they certainly do. With active seasons, you get a much higher probability of landfalling storms of Saffir-Simpson 3-5 categories."

Gray took hurricane activity for the past 98 years and calculated the ratio of the most active 33 years to the least active 33. The ratio was eight to one. He says that simple comparison, and the historical record, support his contention.

"We got so many landfalling intense storms along the U.S. East Coast in the 1940s through the 1960s that the NTC simply doesn’t explain it all," he said. "You also have to take into account the Atlantic Ocean thermohaline circulation."

And, since North Atlantic sea surface temperatures and salinity contents have been rising since 1994, Gray believes the Atlantic conveyor belt may again produce conditions that will generate more intense hurricane landfalls in coming years.

Curiously, the connection between the conveyor belt and Gulf Coast landfalling hurricanes is very weak to non-existent, Gray said. Nonetheless, in an active year, the Gulf Coast remains more prone to landfalls by major hurricanes than in a weak year.

Gray said that while the NTC and Atlantic thermohaline circulation are fundamental features of landfall probability, they do not work in all years.

"Andrew came into South Florida in an inactive year (1992), and Alicia (1983) came into Houston in an otherwise inactive year and did a billion dollars worth of damage," he said. "Those are exceptions to the general rule, however. Most of the landfalling storms this century come in active years. For example, 1985, 1989, 1995 and 1996 were all active years with intense hurricane landfalls by Elena in 1985, Hugo in 1989, Fran in 1995 and Opal in 1996.

"Despite the exceptions, the model does mean something in terms of long-term probability."