by Judith Curry
We anticipate that it may take a decade for the observations to clarify the situation as to whether the hypothesis has predictive power. – Curry et al. 2006
My first substantive post at Climate Etc. was Hurricanes and global warming: 5 years post Katrina. I recall spending two weeks (!) working on that post (its a very good post, if I say so myself). I no longer actively research this topic, although I do keep up with the literature.
In this post, I review the recent observationally-based literature on the topic of climate change and hurricanes, and in particular address the issue of sorting out natural variability from human caused climate change in terms of impacts on any changes in hurricanes; an issue that i raised in my 2006 publication Mixing politics and science in testing the hypothesis that greenhouse warming is causing a global increase in hurricane intensity.
For a quick update, see an an article by CarbonBrief that includes some interviews with hurricane/climate scientists. Also note a blog post by Kerry Emanuel at the CCNF Climate change and Hurricane Katrina: what have we learned?
Let me begin here with an update from some of the antagonists in the 2005/2006 hurricane wars – new papers by Chris Landsea and Phil Klotzbach (protégée of Bill Gray) and Greg Holland.
Landsea and Klotzbach
Phil Klotzbach and Chris Landsea have a new paper in press at J. Climate:
Extremely Intense Hurricanes: Revisiting Webster et al. (2005) after 10 Years
Abstract. Webster et al. (2005) documented a large and significant increase in both the 23 number as well as the percentage of Category 4-5 hurricanes for all global basins from 24 1970-2004, and this manuscript examines if those trends have continued when including 25 ten additional years of data. In contrast to that study, as shown here, the global frequency of Category 4-5 hurricanes has shown a small, insignificant downward trend while the percentage of Category 4-5 hurricanes has shown a small, insignificant upward trend between 1990 and 2014. Accumulated Cyclone Energy globally has experienced a large and significant downward trend during the same period. We conclude that the primary reason for the increase in Category 4-5 hurricanes noted in observational datasets from 1970 to 2004 by Webster et al. was due to observational improvements at the various global tropical cyclone warning centers, primarily in the first two decades of that study.
Their main figure:
Their argument for ignoring data prior to 1990:
The mid to late 1970s heralded the advent of geostationary coverage around much of the world with the launching of GOES-1, 164 Meteosat-1, and GMS-1. However, the North (and South) Indian Ocean lacked direct geostationary satellite data until 1989 with the launching of Meteosat-5 (Knapp and Kossin 2007). Consequently, the archived best track data from JTWC only reports one Category 4-5 hurricane from 1970-1989 for the North Indian Ocean, whereas, during the most recent twenty-year period from 1995-2014, 13 Category 4-5 hurricanes have been observed. Landsea et al. (2006) identified several missed Category 4-5 hurricanes before 1990, which were instead considered to be weaker TCs. After the advent of geostationary satellites, techniques we re developed and refined for interpreting the intensity from satellite imagery. In particular, the Dvorak Technique – a position and intensity pattern recognition scheme – was first developed in the early 1970s for visible imagery (Dvorak 1975) and later revised to include infrared imagery (Dvorak 1984). The adaptation of this tool globally became standard during the 1980s once the Regional Specialized Meteorological Centers were established for monitoring and forecasting tropical cyclones (Velden et al. 2006). In addition, operational responsibility for the northeast Pacific basin shifted from the Weather Service Forecast Office, Redwood City, California, to the National Hurricane Center following the 1987 season, which appears to have caused an artificial jump in 1988 in analyzed intensities between the two agencies (Todd Kimberlain, personal communication, 2015). Finally, the demise of extensive aircraft reconnaissance missions in the Northwest Pacific basin in 1987 likely led to a non negligible impact upon analyzed intensities. The efforts within the IBTrACS project (Knapp et al. 2010) have led to global TC data sets, but, while these data are now easily available, there has yet to be an internationally agreed-upon standardized global TC intensity database.
JC comment: The debate on the increase in % CAT45 hinges on whether the data from 1985-1989 is of useful accuracy. I concur that the global intensity prior to 1980 isn’t useful, but through the 1980’s the data become increasingly reliable. Missed Indian Ocean CAT45 should not have an overwhelming affect on the statistics, since Indian Ocean tropical cyclones constitute only 25% of global TCs. Prior to 1988, Northwest Pacific Basin intensities were arguably MORE accurate (aircraft reconnaissance) – NW Pac tropical cyclones constitute about 40% of global TCs.
Note: here is an email response from Klotzbach specifically regarding 1985-1989:
A couple of primary issues with that pentad are:
1) Aircraft reconnaissance ended in the NW Pacific in 1987. There were probably some issues going purely to satellite data in 1988 that are hard to quantify. JC comment: Quantitatively, potentially how big a deal is this? Worth throwing out 5-10 years of data?
2) Even more problematic were TC classifications in the NE Pacific prior to 1988. The National Weather Service office in Redwood City, CA was responsible for that basin until 1988, and their estimated intensities are quite suspect. For example, correlations between large-scale parameters such as SSTs show very little correlation with ACE over the period from 1971 (when consistent NE Pacific observations began) and 1987, while they show robust correlations with ENSO from 1988-on. JC note: NE Pacific constitutes 17% of global TCs. What is the worst case scenario as to ho much this could impact the global stats?
Well even if you accept PK’s rationale for throwing these data out, there shouldn’t be a problem with adding 1988 and 1989 back into the dataset.
In my previous post, I included this diagram (updated through 2009):
Second point: If the idea is to test a relationship between increasing %CAT45 and increasing SST, then you need to look at the period when the SST is increasing. Starting at 1990, you only see a relatively small increase in SST (well somewhat larger if you are buying the Karl et al. analysis). Every effort should be made to add back the data from the 1980’s, with appropriate corrections and/or error bars.
Holland and Bruyere
Published in Climate Dynamics:
Recent intense hurricane response to global climate change
Abstract. An Anthropogenic Climate Change Index (ACCI) is developed and used to investigate the potential global warming contribution to current tropical cyclone activity. The ACCI is defined as the difference between the means of ensembles of climate simulations with and without anthropogenic gases and aerosols. This index indicates that the bulk of the current anthropogenic warming has occurred in the past four decades, which enables improved confidence in assessing hurricane changes as it removes many of the data issues from previous eras. We find no anthropogenic signal in annual global tropical cyclone or hurricane frequencies. But a strong signal is found in proportions of both weaker and stronger hurricanes: the proportion of Category 4 and 5 hurricanes has increased at a rate of 25–30 % per “C of global warming after accounting for analysis and observing system changes. This has been balanced by a similar decrease in Category 1 and 2 hurricane proportions, leading to development of a distinctly bimodal intensity distribution, with the secondary maximum at Category 4 hurricanes. This global signal is reproduced in all ocean basins. The observed increase in Category 4–5 hurricanes may not continue at the same rate with future global warming. The analysis suggests that following an initial climate increase in intense hurricane proportions a saturation level will be reached beyond which any further global warming will have little effect.
JC comment: This is an interesting analysis, and addresses the attribution issue in a novel way. However, IMO the main challenge to their analysis is the issue of internal variability. Here is what they have to say on this topic:
There is a possibility that some of the observed trend arises from internal tropical cyclone variability that has aliased into the 35-year period used here. However, we consider that a substantial contamination from internal variability is highly unlikely. By using global mean temperatures in the development of the ACCI, we have explicitly excluded all but external forcing factors from this aspect of the analysis. There could be an influence of 11-year sun cycles or impulsive events such as volcanoes, but we suggest that the period chosen is too long for these to have a cumulative effect on the trend.
For global hurricane proportions, we are aware of no association with internal variability such as the Pacific Decadal Oscillation or similar. The consistent regional Cat 4–5 relationshipwith globalACCI argues for the regional changes observed here being unaffected by relative processes and thus due largely to global changes.
I am unconvinced that natural internal variability is a non-issue over the past 35 years.
To address the issue of natural variability, see these plots from tweets by Ryan Maue.
Global Accumulated Cyclone Energy (Maue):
Emanuel’s Potential Dissipation Index for North Atlantic (Maue)
2015 – a record breaking year
If you mostly pay attention to N. Atlantic hurricanes, you may have missed the records being broken elsewhere around the globe. These records are summed up by the following tweets from Phil Klotzbach:
Northern Hemisphere Cat 4-5 hurricanes by Aug 16 (Klotzbach)
Northern Hemisphere ACE through Aug 24 (Klotzbach)
Most of the action is in the NW Pacific; Phil Klotzbach has blogged at Weather Underground on the records being set by the 2015 NW Pacific TC season. The North Atlantic is having a quiet year so far, as predicted (El Nino).
Tweets from Phil Klotzbach:
Northern Hemisphere ACE remains at record levels (400 ACE) thru 8/19, ~90 units ahead of 2nd place (310 ACE) (1971).
NW Pacific ACE at record high levels (257) thru 8/17 – ~30 ACE greater than 2nd place (2002).
Historic central/eastern Pacific outbreak- 3 major hurricanes at once for the first time on record!
Hurricane Jimena is the 13th Cat. 4-5 TC of the 2015 N Hemis season, 4 more than any other year to date since ’71
Thru 8/27, the average 1981-2010 NW Pacific season had 3.3 Cat 3+ typhoons. In 2015, the NW Pacific has already had 9 (273% of normal!).
N Hemis ACE remains at record levels thru Aug 24 (435 ACE) – over 100 ACE units ahead of 2nd place (1971 – 321 ACE).
75% of all hurricanes forming in the Northern Hemis in ’15 have reached Cat. 4-5, breaking record set in 2007 (64%).
JC comment: Looks like 2015 will significantly change the statistics of the assessment of %CAT45.
Chris Landsea on U.S. landfalls
Chris Landsea recently published a comment that has a useful updated plot of U.S. landfalls.
The figure shows that there has been a small, statistically insignificant downward trend in the frequency of U.S. [landfalling] hurricanes in this century-long time series. Instead, the record is dominated by Interannual to decadal scale variability, with the busies periods occurring in the 1910s, the 1930s to the 1950s, the mid-1980s, and the mid 2000s, while the quietest periods are seen during the 1920s, the 1970s to the early 1980s, the early 1990s, around 2000 and the last few years.
It seems that the principal scientists involved in this research are saying pretty much the same thing they were each saying 10 years ago.10 additional years of data have helped a little bit in sorting all this out, but losing 1-2 decades of data on the front end of the data series definitely hinders our ability to settle the disagreements over this.
SST is hypothesized to play a role in intensity, NOT frequency. Hence metrics such as ACE and PDI, while interesting in their own right, don’t really address the issue of intensity directly. The metric %CAT45 seems to have held up as the metric that is increasing, although the amount of the increase depends on what period you consider – starting the time series at 1990 misses much of the late century temperature increase and the apparent increase in %CAT45 during the 1980’s.
The issue of natural variability seems perhaps overemphasized by Landsea and underemphasized by Holland. In any event, how to sort out the natural from anthropogenic causes of the TC variability remains the key issue in all this. In Curry et al. (2006), I wrote:
A number of natural internal oscillations of the atmosphere-ocean system have a large impact on SST (e.g. El Nino, North Atlantic Oscillation). However, decadal-scale oscillations tend to be specific to each ocean basin and are often anti correlated from one basin to another.
Since this paper was published, we now understand much more about the natural models of variability on decadal to multi-decadal time scales. These oscillations (e.g. the AMO, PDO, NGPO, etc) influence tropical cyclone activity through organizing circulation systems in the atmosphere and ocean that influence SST, wind shear, atmospheric humidity, etc. Most significantly, a number of these oscillations are synchronized and interact in a coherent way (stadium wave). Hence, a coherent multidecadal global signal in tropical cyclone activity associated with the stadium wave would not be a surprise. Unfortunately, 25 years (or even 45 years) of data is not sufficient to identify and understand such a signal.
So apart from the challenges of detecting an increase in %NCAT45, the issue of attributing any increase to natural variability versus human caused global warming remains outstanding. Both are almost certainly contributing to any increase; the question is of course whether human caused warming dominates over natural variability. Given the large interannual and interdecadal variability (and the relatively short data record, it isn’t possible with our current understanding to tease out the fractional attribution. The method proposed by Holland and Gruyere is interesting, but it somehow needs to better account for natural internal variability.
So . . . another 10 years, what will that bring? Well, this time I won’t predict that this will be sorted out in another 10 years.