In all low latitude ocean basins except the South Atlantic and Southeastern Pacific, tropical cyclones develop during the warm seasons and pose hazards to shipping and coastal areas as they move slowly westward and poleward around the massive subtropical high pressure systems. The strongest of these storms are called hurricanes in the North Atlantic and East Pacific, typhoons in the western North Pacific, and cyclones in the South Pacific and Indian Oceans. For tropical storms to form, they require (1) warm SSTs (26˚C (79˚F) or higher), (2) weak upper level winds and wind shears, preferably with diverging flow aloft, (3) a low level cyclonic (counterclockwise) circulation, and (4) atmospheric instability that leads to organized convective clouds (thunderstorms). We will look at evidence regarding any recent change in frequency or intensity for these storms, and then discuss model projections for the latter part of the 21st century.
Tropical Storm Frequency
First, let’s look at the frequency of storms in the North Atlantic Basin. Figure 1 shows the trend of Atlantic tropical storms from 1878 to 2006.  Prior to the advent of meteorological satellites in 1965, the number of storms would have been significantly underestimated without some correction, since the only oceanic observations came from sparse ship reports. The trend line was adjusted to estimate the number of missing storms, shown by the blue curve at the bottom of the figure. The resulting trend shows a slight average increase during that period (about +1.6 storms per century), although there were two minima during the 129 year period. The number of hurricanes from 1878 to 2006 seems to correlate with the observed increase in both global mean temperature and sea surface temperature (SST) (Figure 2).  However, there is no statistical significance to the trends in the other three curves: adjusted hurricane counts, U. S. landfalling hurricanes, and Atlantic SST relative to Tropical SST. Much of any increase in storm frequency has been determined to be due to increases in short duration (<2 day) storms that would likely have gone undetected in the early period of record.  Thus, there is much uncertainty in the trends in the Atlantic Basin, due mainly to the relatively short historical period of quality observations. The correlation is even weaker when we consider landfalling U.S. Atlantic hurricanes. Globally, the results are similar. A WMO report indicated that there was uncertainty that past trends in tropical storm occurrence exceeded the natural variability. 
Figure 1.Number of Atlantic Ocean Tropical and Subtropical storms from 1878 to 2006 adjusted prior to 1964 to account for undetected storms in the pre-satellite era (blue curve at bottom). (Source: National Hurricane Center (NOAA) HURDAT best track dataset)
Figure 2. Normalized trends of Tropical Atlantic indices for the period 1878-2011. Dashed blue lines for the top three panels show statistically significant trends. (adapted from Vecchi and Knutson, 2011)
Projections of tropical storm frequency through the 21st century were obtained using a High Resolution (50 km grid) Atmospheric Model (HiRAM) that incorporates projected changes in SST . The model was downscaled into the Geophysical Fluid Dynamics Laboratory (GFDL) Hurricane Model following storm genesis to determine tracks and intensities. The results show a decrease in frequency of tropical storms globally by 6-34%. The only region showing an increase is the Eastern North Pacific (from the west coast of Mexico to Hawaii) (Figure 3). This seems counter-intuitive since SSTs are forecast to rise significantly, so we would expect tropical storm frequency to increase also. Possible reasons for the decrease are: (1) a weakening of tropical circulations that lead to the required thunderstorm activity, or (2) drier mid-tropospheric conditions which would also suppress convection.
Figure 3. Frequency of tropical storms projected for the late 21st century based on anticipated global warming and resulting SST increases. The difference (late 21st century minus present-day) is shown in the bottom panel. Except for the Eastern North Pacific, tropical cyclone frequency is expected to decrease. (From Knutson et al., 2015)
- Tropical Storm Intensity
The intensity of tropical storms and hurricanes is measured using the maximum sustained wind speed, minimum sea level pressure (SLP), and something called the Power Dissipation Index (PDI) which combines cyclone strength, duration, and frequency. A comparison of the PDI for North Atlantic cyclones with concurrent SST is shown in Figure 4.  The PDI decreased to a minimum in the period from 1970-1980. Since that time, an uptick has occurred, followed by another decrease after 2005. It correlates pretty well with the trend of SST, except for the period 2008-2013.
Figure 4. Observed trend of the five-year mean Power Dissipation Index (PDI), a measure of tropical storm intensity, versus SST from 1951-2013. (2016 Update to data from Emmanuel, 2007 )
Tropical storms and hurricanes intensify when they move into areas of favorable conditions such as: (1) very warm SSTs that extend through deep layers of the ocean, and/or (2) weak winds aloft with pronounced divergent outflow. Unfortunately, short range prediction of intensification is poor compared to prediction of storm tracks. Long range climatic predictions of storm intensities must be based on assumptions about the model-simulated life cycles of storms under future climatic conditions.
Projections of the future intensity of tropical storms assuming greenhouse gas warming show increases of 2-11% by 2100. Coupled with this is an expected increase in the precipitation rate of about 20% within 100 km of the storm centers.  Considerable variations were observed regionally from basin to basin. As an example, Figure 5 shows the expected distribution of minimum Sea Level Pressure (SLP) for a scenario in which CO2 levels increase 1% per year for 80 years (heavy line) versus the current mean SLP distribution (thin line with open circles). The mean minimum SLP is about 10 millibars lower for the high CO2 scenario. More disturbingly, recent research suggests that if global warming occurs as forecast, extremely rapid intensification (>60 knots in 24 hours) of landfalling hurricanes in the United States could occur once every 5-10 years as opposed to once every 100 years in the past.
Figure 5. Simulated distribution of hurricane minimum Sea Level Pressure (SLP) for a high CO2 scenario possible after 80 years (thick line) versus current distribution.
In summary, recent trends in the frequency and intensity of tropical cyclones have not been significant enough to be certain that they exceed the natural variability, since the historical record is so short. Projections of future storm activity through the end of the 21st century based on numerical models portend a significant reduction in storm frequency but a likely increase in intensity, assuming further increases in SST due to higher levels of greenhouse gases. There is also the possibility that there will be more frequent occurrences of very rapid intensification prior to landfall in the United States, mostly along the Gulf of Mexico coastline. All this of course, is predicated on the continued rise of global mean air and sea surface temperatures. Although this is likely, the magnitude of the increase is not a certainty by any means.
- Vecchi, G. and T. Knutson, 2008: On Estimates of Historical North Atlantic Tropical Cylcone Activity. Journal of Climate, 21, pages 3580-3600.
- ___________________, 2011: Estimating Annual Numbers of Atlantic Hurricanes Missing from the HURDAT Database (1878-1965) Using Ship Track Density, Journal of Climate, 24, pages 1736-1746.
- Landea, C., G. Vecchi, L. Bengtsson, and T. Knutson, 2010: Impact of Duration Thresholds on Atlantic Tropical Cyclone Counts. Journal of Climate 23, pages 2508-2519.
- Knutson, T., J. McBride, J. Chan, K. Emmanuel, G. Holland, C. Lansea, I. Held, J. Kossin, A. Srivastava, and M. Sugi, 2010. Tropical cyclones and climate change. Nature Geoscience, 3, pages 157-173.
- _______, J. Sirutis, M. Zhao, R. Tuleya, M. Bender, G. Vecchi, G. Villarini, and D. Chavas, 2015. Global Projections of Intense Tropical Cyclone Activity for the Late Twenty-First Century from Dynamical Downscaling of CMIP5/RCP4.5 Scenarios. Journal of Climate, 28, pages 7203-
- Emanuel, K.A. 2016 update to data originally published in: Emanuel, K.A. 2007. Environmental factors affecting tropical cyclone power dissipation. of Climate Vol. 20(22), pages 5497–5509.
- Knutson, T., and R. Tuleya, 2004: Impact of CO2-Induced Warming on Simulated Hurricane Intensity and Precipitation: Sensitivity to the Choice of Climate Model and Convective Parameterization. Journal of Climate, 17, pages 3477-3495.
- Emmanuel, K, 2017: Will Global Warming Make Hurricane Forecasting More Difficult? Bulletin of the American Meteorological Society, Vol. 98, March, pages 495-