Fall in Tropical Cyclone Activity in Australia

Scientists use stalagmites to show that today’s tropical cyclone activity is lowest in the past 550 to 1500 years

Sage Vals congratulates the authors, but finds some of their claims a little bit stretched

In a letter in Nature this week (30 January 2014), Jordahna Haig, Jonathan Nott and Gert-Jan Reichart reported their achievement in calibrating the 18O content of stalagmite growth bands with the recent observational record of tropical storm activities in two locations in Australia (1). Using this calibration, they were able to discover tropical storm activity in parts of Australia dating back as far as 1500 years


It is established that rain from tropical cyclones is depleted in the heavier oxygen isotope 18O relative to monsoonal rainfall. How much the water is depleted at any given site varies with the intensity of the storm, the duration of the storm, and the distance of the site from the centre of the storm.

The rainwater is incorporated in the precipitate that forms stalagmites. The stalagmites used in this study are seasonally banded, meaning that it is possible to identify annual growth rings. By measuring the isotopic content of each stalagmite growth ring an indicator of storm activity each year is obtained. Two stalagmites, one from Western Australia and another from Queensland, were used in this study.

A New Storm Index

The problem is how to relate the 18O content to actual storm activity. The authors propose “a new tropical cyclone activity index (CAI)” This is based on actual observations of tropical storms, using a formula which relates a function of storm wind speed, storm radius, duration of storm and the distance of the storm centre from the site of the stalagmite for each storm, averaged over the number of tropical cyclones in a season. Observational data for the period from 1990 to 2010 was used, and the results compared to the 18O content of the stalagmites for the appropriate years.

The CAI, as formulated, correlates well with the 18O content for the 20 years used, allowing the stalagmite record to be calibrated against observational data. Furthermore, the 18O content of the stalagmites for the years 1970 to 1990 matches well with observational data (which is less detailed than the data used for calibration) for that period too.

The authors conclude that the now-calibrated stalagmite data can be used to estimate tropical storm activity for as far back as the stalagmite records allow. Using this insight, the CAI index of tropical storm activity for the last 1500 years in the case of the Western Australia stalagmite, and 700 years for the Queensland stalagmite, was derived. The graphs of the results can be seen here, http://www.nature.com/nature/journal/v505/n7485/full/nature12882.html (fig 3) along with the abstract and other figures (and the full paper £).

The authors observe some cyclical changes in tropical storm activity over the period, but detect a break in that cyclicity in more recent years in terms of a significant decline in activity measured by the CAI. This they conclude is evidence of anthropogenic climate change, and their results accord in quality, although not in timescale, with some modelled climate change projections

Significant Achievement

The use of the isotopic composition of stalagmite precipitates in determining a high (annual) resolution storm activity record is significant. As the authors recognise, their research needs to be replicated with more stalagmite samples from other areas of the world in order to eliminate the possibility of coincidental correlation of the data. Should the CAI method prove robust, it could prove very useful in climate, and potentially paleoclimate studies. It would always be necessary to show that any stalagmites used had not been subject to alteration since their formation, but subject to this qualification, the method should have applications in reconstructing storm activity, possibly over millennial or even longer timescales.

Issues and Criticisms

There is a problem with the CAI index. While it can obviously give a measure of overall storm activity, it is weighted towards storms producing high winds. Without getting too deeply into the mathematics of the CAI, the function uses the cube of the observed maximum wind strength, but only the first powers of the radius of the centre of the storm, the duration of the storm and distance from the relevant stalagmite. Thus, high wind speed storms are heavily weighted in the index. This might be of importance in terms of destructive potential, but may mean that some climate changes are less detectable. For example, a reduction in the number of very high wind speed storms may totally obscure an increase in storm frequency. It is appreciated that the index has to reflect the way that the stalagmite picks up the 18O from the rain precipitated, but it does mean that some climate changes may go undetected. Because the stalagmite record is high resolution doesn’t make the corresponding weather conditions highly resolved.

The authors’ claim “that it is now possible to discern natural variability from anthropogenically induced change”. This seems to be based on the apparent post-industrial period change in the frequency of tropical storms seen in the high-resolution data. This variation being unlike the observed cyclical variations in the pre-industrial period. While this conclusion cannot be dismissed out of hand, it is an assumption. Having eliminated known cyclical explanations, the authors take for granted that no other explanation is possible. They do not, it should be noted, make any claims as to the mechanism by which human activity has caused this effect. Other workers have been more circumspect, for example Callaghan & Power, 2011 (2), as cited by the authors of this letter.

However, the authors make greater claims than this. In particular, they suggest that a method such as this might be used “potentially to forecast future trends in cyclone activity under changing climate conditions”. I am not sure how this can work. They have created a high resolution technique for measuring past and (almost) current tropical cyclone activity, but how does a measuring tool make forecasts? I could invent a very high resolution thermometer, which would measure temperatures to the millionth of a degree, but this would never tell me why the temperatures I measured were changing: just that they were. It would not tell me what changes might occur in future. I could perhaps see patterns in the measurements over time, but these could only be used to make predictions if I assumed that the present was conformable to the past. By definition, and by the authors’ own methodology, it is discerning breaks in conformity that the method is used for, not to predict future further breaks. The best that can be hoped for is these results could be used to assist in the construction of better climate models which could be calibrated against past tropical storm records (i.e. backcast) before being used to forecast future climates.

Despite the immoderate claims made for it, the research reported in this paper is excellent work. The method devised should prove to be a great tool for future research on past climates and can be expected to be of much value.


1. Australian tropical cyclone activity lower than at any time over past 550-1,500 years. Haig, Jordahna, Nott, Jonathan and Reichart, Gert-Jan. 7485, 2014, Nature, Vol. 505, pp. 667–671.

2. Variability and decline in the number of severe tropical cyclones making land-fall over eastern Australia since the late nineteenth century. Callaghan, Jeff and Power, Scott B. 3-4, 2011, Climate Dynamics, Vol. 37, pp. 647-662.