What is the impact of climate change and subsequent increase in sea surface temperature having on tropical hurricanes? A critical evaluation

With the continuing population increase in coastal areas, predicting fluctuations in tropical hurricane activity is of obvious importance to society and scientists alike. Concerns about the possible effects of global warming on tropical hurricane activity have motivated a number of theoretical, modelling and empirical studies, in the hopes that understanding the resulting trends will allow for the projection of how tropical storms will differ under various scenarios of climate change. Understanding how hurricane intensity is impacted allows for the potential mitigation, or at least preparation, for future natural disasters.

Following predictions about transient greenhouse gas changes made in the 1980s, the global surface temperature has increased by ≈0.2°C per decade (Hansen et al, 2006). Similarly, sea surface temperature has followed an upward trend over the last 50 years, with increases on average of 0.5°c between 1970 and 2004 (Elsner and Kocher, 2004). With current predictions seeing a continued increase in both surface and sea temperature, is it important we understand how these changes may impact tropical hurricane formation.

Our basic understanding of how tropical hurricanes form suggests that there could be a relationship between hurricane activity and rising sea surface temperatures (SST). It is widely acknowledged that SST >36°c is a requirement for tropical hurricane formation (Gray, 1968) and as SSTs are set to rise further, some tropical ocean basins may face an increasing number of more intense tropical hurricanes (Chu and Clark, 1999). This is because the higher SST and resulting higher specific humidity can contribute more energy during storm formation, allowing for more destructive storms (Webster et al, 2005).

There is some evidence that tropical hurricanes have already been impacted (Elsner et al, 2008; Kossin et al, 2014) and common consensus has been that future projections for climate change indicate that anthropogenic warming will at least cause globally averaged intensity of tropical hurricanes to increase (Knutson et al, 2010) however, there are also some who believe the links between SST and hurricane intensity are weak and explained by other factors.

Whether the characteristics of tropical hurricanes have changed, or will change in a warming climate, has been the subject of considerable investigation, often with conflicting results (Knutson et al, 2010). This difference in findings stems from the fact that trend detection is hindered by substantial limitations in the quality and availability of tropical hurricane-global historical records (Knutson et al, 2010). The most comprehensive datasets were not primarily intended as datasets to be used for trend analysis and despite using the best techniques available (Walsh et al, 2015) the density of reporting ship traffic was relatively sparse during the early decades of the records, such that if storms from the modern era (post 1965) had hypothetically occurred during those earlier decades, a substantial number would not have been directly observed by the ship-based ‘observing network of opportunity’ (GFDL, 2017). After adjusting models for the missing storms, one study (Vecchi et al, 2008) found that the rising trend in tropical hurricane storm count and intensity was no longer statistically significant and only a small nominally positive upward trend from 1878-2006 was observed (GFDL, 2017). Similarly, a global analysis of tropical hurricane intensity trends from 1981-2006 found increases in the intensities of the strongest hurricanes (Knutson et al, 2010) however, these results are again hindered by the short time period of the dataset.

Whilst some models show significant trends, the problems revolving around the current available dataset due to limited ship density in the pre-satellite era and the resulting missing storms mean scientists are still uncertain whether any changes in the past tropical hurricane activity can be attributed to climate change and rising SST or errors in the modelling and natural climate variability (Knutson et al, 2010).

Tropical hurricane projections rely heavily on dynamical models including global climate models, higher resolution global atmospheric models forced by SSTs from global climate models or even higher resolution regional downscaling models (Knutson et al, 2010), each method has strengths and weaknesses and when applied to the same model they can produce wildly different results (Tory et al, 2014). This causes a lack of consensus among the researchers. For example, two statistical models of hurricane activity vs SST, both of which performed comparably during the historical period, give dramatically different projections for late 21^st century activity, with one predicting an increase of 300% in power dissipation by 2100 (Knutson et al, 2010), the other projecting a much smaller change which falls more in line with other dynamical models (Knutson et al, 2010). Other research produced by Bender et al. (2010) projected a significant increase (+90%) in the frequency of very intense (category 4 and 5) hurricanes when using the CMIP3/A1B 18-model average climate change projection. However, subsequent downscaled projections run by Knutson et al. (2013), whilst still showing an increase in category 4 and 5 hurricanes, was only marginally significant for the early 21^st century (+45%) and the last 21^st century (+39%) (GFDL, 2017).

The examples provided, whilst simplified for this paper, illustrate how different dynamical, statistical or downscaling techniques can offer substantially different projections in the response to tropical hurricanes and rising SST. In terms of general modelling approaches, all techniques can provide complimentary approaches and are worth pursuing, however each has its limitations and the results given must be interpreted with caution.

Large natural fluctuatio