Joint modeling of hurricane numbers and major hurricanes in different basins
Hurricanes are serious meteorological phenomena that can cause massive material damage and there is some evidence that hurricane numbers and speeds are changing due to manmade climate change. Therefore it is important to have reliable models for both the numbers and speeds of hurricanes. We model the marginal distribution of the number of hurricanes per year in a given basin as a Poisson regression depending on climatic inputs such as ENSO (El Niño / La Niña) and SST (sea surface temperature). Then, the correlation between the hurricane numbers in different basins is modeled using copulas. Our evidence backs up empirical evidence of a negative relationship between the hurricane numbers in the North Atlantic and East Pacific basins. Secondly, conditional on the hurricane numbers, we model wind speeds independently using lognormal regression structures. Our results suggest that there is some evidence that more hurricanes reach land now than previously.
Palabras clave: Hurricanes SST ENSO generalized linear models copulas