Using Three Global Climate Indices to Forecast Hurricane Activity in the Pacific and Atlantic Oceans

Abstract:

Quantitative relationships between global climate indices and hurricane activity in the Pacific and Atlantic Oceans have not been widely studied. A few studies have explored qualitative relationships between hurricane activity and such climate indices as the North Atlantic Oscillation and sea-surface temperatures, among others. The current work presents the most comprehensive analysis of the potential relationships between 39 different climate indices and hurricane activity using regression and frequency analysis. Attempts are made to develop statistical relationships between any one of these indices and hurricane activity in the eastern and western Pacific as well as the Atlantic Oceans. There were three climate indices, one per region, showing significantly higher correlation in each region. They were the ENSO Precipitation Index (EPI) in the western Pacific, the Atlantic Multi-decadal Oscillation (AMO) in the eastern Pacific, and the Atlantic Meridional Mode (AMM) in the Atlantic. The linear relationships between each index and hurricane numbers resulted in Pearson-R values of near 0.65 or greater. In addition, the Madden-Julian Oscillation showed some correlation with hurricane activity in each region and therefore was included in the analysis. Several important results were found during these analyses. For instance, the relationship between the AMM index and hurricane numbers in the Atlantic Ocean revealed that the average July – October AMM index was greater than -0.5 within a range of -5.0 to 5.0 for years within the last 70 years when the number of hurricanes during that same period was greater than 7. It is also shown that the number of hurricanes expected to be exceeded or not exceeded at frequencies of 50- to 100-years, for example, varies substantially depending on the range of AMM index values being analyzed. Similar results are shown for the eastern and western Pacific Ocean as well. Such relationships provide forecasters with a simple tool using only one variable that provides a rough estimate of hurricane activity for the upcoming hurricane season.