Insurers and underwriters are increasing their Hurricane Ike loss estimates by tens of millions of dollars, highlighting the challenges of catastrophe models for the offshore energy sector, industry observers say.

Advent Capital (Holdings) P.L.C. this month increased its Ike loss estimates by about $39 million, and Platinum Underwriters Holdings upped its loss estimates by about $60 million last week.

Catastrophe models also underestimated losses from Hurricanes Katrina and Rita in 2005.

Experts say predicting storm damage to oil rigs, production platforms and other offshore facilities is more difficult than cat modeling for refineries and onshore energy structures. One reason is that most offshore modeling began after Hurricane Ivan in 2004, said Tommy Laurendine, assistant vp-U.S. exploration and production at Liberty International Underwriters in Houston.

"It's really in its infancy," Mr. Laurendine said of offshore modeling. "Quite frankly, we only have three storms that the modelers can use to test their models against" because 2008 hurricane data is not yet available, he said.

While hurricane and earthquake damage to onshore facilities often is made public, offshore damage typically has been guarded closely by individual companies, said Mr. Laurendine, who worked at an oil company and for the U.S. Minerals Management Service prior to Liberty International.

Also, the decline in oil production in the 1980s caused many industry firms to cut back on hiring and training offshore structural engineers, who now are outnumbered by onshore structural engineers, he said.

Having capable engineers is important, he and others say, because modeling offshore damage is far more complicated than onshore losses.

"Ocean, current, wind action--none of these things are very predictable," said Steve Maloney, senior consultant in the enterprise risk management practice at Stamford, Conn.-based Towers Perrin. "The combination can be reinforcing or canceling."

Offshore models predict the wind created by the hurricane, the waves created by the wind and the structural damage caused by the waves. The relationship between those variables is not linear, Mr. Laurendine said.

"What causes the destruction is waves, not wind," he said. "The problem is it's much easier to teach wind-force calculations than it is to teach wave-force calculations."