In response to some ongoing debate in the hurricane research community, I decided to perform a very simplistic crunching of numbers. Some recent studies (e.g. by Webster, Curry, Holland, and others) claim that the number of very intense hurricanes -- category 4 and 5s -- has increased two-fold over the past 30-50 years, whereas others (e.g. Landsea, Klotzbach) claim that no such trend exists and that the available storm databases are of too poor quality to draw any definitive conclusions. Each of these studies has looked purely at the numbers, with no break down by other modulating atmospheric factors.

So, based upon conversations with a colleague, I decided to split out hurricane seasons based upon ENSO phase. A list of years of ENSO phase may be found here. I composited the seasons for each phase and looked at the following statistics: number of storms (NS), number of hurricanes (H), number of intense hurricanes (IH), number of category 4 & 5 hurricanes (N45), and season-long duration of the cat. 4 & 5 hurricanes (hrs C45). I looked only at the Atlantic, given that it is where much of the focus in this debate lies and that it has the best quality storm database, and only at the years from 1970-2005 (e.g. in the satellite and reconnaisance era). The years studied largely match those of Klotzbach's recent work.

Note: by no means is this a scientific-quality study. I've taken efforts to ensure the validity of the results, but the sample sizes are relatively low for each composite of seasons and a complete study would also include the East and Northwest Pacific basins. Statistical tests on whether each of the averages are different from each other would help alleviate the relatively small number of cases.

With that said, here are some results. All seasonal averages exclude the minimum and maximum values of each statistic so as to not have the statistics swayed by extreme seasons in either direction.

NEUTRAL YEARS: 11.35 NS, 6.29 H, 2.47 IH, 1.29 N45, 54.35hr C45
minimum values -- 6 NS (1981 and 1985), 2 H (1981), 0 IH (1985 and 1993), 0 N45, 0hr C45 (both 5 seasons)
maximum values -- 28 NS, 15 H, 7 IH, 5 N45 (all 2005), 308hr C45 (2004)

LA NINA YEARS: 11.00 NH, 6.17 H, 2.17 IH, 1.17 N45, 37.00hr C45
minimum values -- 7 NS, 3 H, 0 IH (all 1972), 0 N45, 0hr C45 (both 1972 and 1975)
maximum values -- 15 NS (2000), 8 H (1999 and 2000), 5 IH, 5 N45, 210hr C45 (all 1999)

EL NINO YEARS: 9.14 NS, 4.71 H, 1.71 IH, 0.57 N45, 44.57hr C45
minimum values -- 4 NS (1983), 3 H (1983 and 1987), 1 IH (5 seasons), 0 N45, 0hr C45 (both 4 seasons)
maximum values -- 16 NS (2003), 10 H (1998), 3 IH (3 seasons), 3 N45 (1988), 228hr C45 (2003)

So, what does this all tell us? Neutral years seem to see the most activity, followed by La Nina years and El Nino years. The differences between neutral and El Nino years are 2.21 NS, 1.58 H, 0.76 IH, 0.72 N45, and 9.78hr C45. On first glance, these would appear to be significant differences, but small sample size issues exist. A quick t-test on the intense hurricane data suggest the difference is significant to the 72% confidence level -- not all that high. The difference in named storms is significant to the 81% confidence level -- a bit better, but still not that high. More years and/or adding in other basins would help establish greater confidence.

Since 1994, all but 2 seasons -- 1997 and 2002 -- have been classified as either La Nina or neutral seasons. Given that these seasons, as a whole, see greater activity than do the El Nino seasons, this could explain part of what we have seen over the past 10 years in this basin. All of the recent seasons in all three groups have seen greater activity than their means for the years between 1970-1995, so performing these analyses with an independent set of data (e.g. not including 1995-2005) would likely not change the results significantly. The question thus becomes one of what has caused the reduced frequency of El Nino years relative to all other years. The multidecadal signal in the basin is purportedly one that reduces the frequency of El Nino events, though the reasons behind this are not well understood. In fact, the reasons why El Nino events would be suppressed in the first place independent of the multidecadal signal are not well understood at all.

Could it be a global climate change signal? Possibly, but the skeptic in me wants to see more data before jumping on that bandwagon. These preliminary numbers suggest that it is likely not an anthropogenic climate change signal, but the questions noted above remain. I just have a hard time believing that we would suddenly see a large shift in activity -- specifically that of intense hurricanes -- solely based upon human-induced global climate change. The authors of those studies state that when you look at all major hurricanes, not just the cat. 4 & 5 storms, the trend does not exist. Could this be due to improved detection of intense storms across the globe? Perhaps.

The overall point: there's still a lot we don't know. But, the more work we do, the better we can hope to understand what exactly is going on with hurricanes and climate change.