by Judith Curry
Final installment in my series on hurricanes and climate change.
7. 21st century projections
The effect of climate change on hurricanes has been a controversial scientific issue for the past several decades. Improvements in the capabilities of climate models, the main tool used to predict future climate, have enabled more credible simulation of the present-day climatology of hurricanes (Walsh et al 2016). The increasing ability of climate models to predict the interannual variability of hurricanes in various regions of the globe indicates that they are capturing some of the essential physical relationships governing the links between climate and hurricanes.
This Chapter addresses climate model projections of hurricane activity out to 2100, in response to manmade global warming. Also addressed is the role of natural modes of climate variability in influencing hurricane activity out to 2050.
7.1 Climate model projections
Apart from the difficulty of simulating hurricane activity in climate models; there is substantial uncertainty associated with climate model projections of 21st century climate change, including the changes in sea surface temperatures and ocean and atmospheric circulation patterns that would cause any changes in hurricane activity. Curry (2018a; Sections 5.1, 5.6) provides an analysis of these uncertainties; a summary of that analysis is provided here.
The climate model simulations of 21st century climate referenced in the IPCC AR5 are based on more than 30 different global climate models from international climate modeling groups. The climate models simulate changes based on a set of scenarios of manmade forcings from changing atmospheric composition, primarily from fossil fuel emissions. ‘Radiative forcing’ is the difference between insolation (sunlight) absorbed by the Earth and the energy radiated by the Earth and its atmosphere back to space. Radiative forcings are influences that cause changes to Earth’s climate system by altering the Earth’s radiative equilibrium, forcing temperatures to rise or fall.
A new set of emissions scenarios, the Representative Concentration Pathways (RCPs), was used for the climate model simulations in the IPCC AR5. In all RCPs, atmospheric CO2 concentrations are higher in 2100 relative to present day as a result of a further increase of cumulative emissions of CO2 to the atmosphere during the 21st century. The four RCPs are named according to radiative forcing target level for 2100. The radiative forcing estimates are based on the forcing of greenhouse gases and other forcing agents. The four selected RCPs include one mitigation scenario that leads to a very low forcing level (RCP2.6), two medium stabilization scenarios (RCP4.5/RCP6) and one very high emission scenario (RCP8.5).
RCP8.5 is sometimes referred to as a ‘business as usual’ scenario. It is not. Rather, it is an extreme scenario that may be impossible. Ritchie and Dowlatabadi (2017) recommend that RCP8.5 should not be used as a benchmark for future scientific research or policy studies.
Table 7.1 summarizes the IPCC AR5 temperature and sea level rise projections for 2046-2065 and 2081-2100. Eliminating RCP8.5 from further consideration here, the likely range of temperature increase by the end of the 21st century is 0.3 to 3.1oC [0.5 to 5.5oF].
Table 7.1 Projected change in global mean surface air temperature and global mean sea level rise for the mid- and late 21st century relative to the reference period of 1986-2005. [IPCC AR5 WGI]
Climate change projections for the 21st century are only as valid as the climate model simulations upon which they are based. Chapters 11 and 12 of the IPCC AR5 describes uncertainties in the climate model-based projections:
“Projections of future states of the global climate are subject to several sources of uncertainty. The first source of uncertainty arises from natural internal variability, which is intrinsic to the climate system, and includes phenomena such as variability in the mid-latitude storm tracks and the ENSO. The existence of internal variability places fundamental limits on the precision with which future climate variables can be projected. The second is uncertainty concerning the past, present and future forcing of the climate system by natural and anthropogenic forcing agents such as greenhouse gases, aerosols, solar forcing and land use change. The third is uncertainty related to the response of the climate system to the specified forcing agents, which is referred to as the ‘climate sensitivity.”
“Simplifications and the interactions between parameterized and resolved processes induce ‘errors’ in models, which can have a leading-order impact on projections. Also, current models may exclude some processes that could turn out to be important for projections) or produce a common error in the representation of a particular process.”
The IPCC AR4 (2007) made the following projection for near-term warming:
“For the next two decades, a warming of about 0.2°C per decade is projected.”
Figure 7.2 provides an update of Figure 11.25 from the IPCC AR5. It is seen that the observed temperatures between 2000-2012 are at the bottom of the envelope of climate model simulations (this period is often referred to as the ‘warming hiatus’). The red hatching in Fig. 11.25 (Figure 5.2) reflects the judgment by the AR5 authors that lowers the projected warming out to 2035 relative to the climate model simulations.
The large El Niño of 2016 has returned the observed temperature curve to near the middle of the envelope of climate model simulations; however the previous large El Niño of 1998 was at the top of the envelope of climate model simulations. The recent data since 2012 continues to indicate that the sensitivity of at least some of the climate models to CO2 forcing is too high.
Figure 7.2 Synthesis of near-term projections of global mean surface air temperature anomalies. Projections from climate models showing the 5 to 95% range using a reference period of 1986–2005 (light grey shade). The maximum and minimum values from climate models using all ensemble members and the 1986–2005 reference period are shown by the grey lines. Black lines show annual mean observational estimates. The red-hatched region shows the indicative likely range for annual mean GMST during the period 2016–2035. [following IPCC AR5 WG I Figure 11.25; updated by Hawkins 2018]. Added green line between 1998 and 2016 reflects the trend between two strong El Niño years.
A key issue is the uncertainty of sensitivity of climate models to CO2. The equilibrium climate sensitivity (ECS) is a measure of the climate system response to sustained radiative forcing, defined as the amount of warming in response to a doubling of atmospheric CO2.
For over thirty years, climate scientists have presented a likely range for ECS that has hardly changed – the ECS range of 1.5−4.5 oC in 1979 (Charney et al. 1979) is unchanged in the 2013 IPCC AR5. While previous assessments have provided a ‘best estimate’ of 3.0 oC, the AR5 did not provide a best estimate value for ECS, stating:
“No best estimate for equilibrium climate sensitivity can now be given because of a lack of agreement on values across assessed lines of evidence.”
At the heart of the uncertainty surrounding the values of ECS is the substantial difference between values derived from global climate models versus values derived from changes over the historical instrumental data record using global energy budget analyses. The median ECS given in IPCC AR5 for global climate models was 3.2 oC, versus 2.0 oC for the median values from historical-period energy budget based studies.
Subsequent to the IPCC AR5, Lewis and Curry (2015) used an observationally-based energy budget methodology with the AR5’s global forcing and heat content estimate time series to derive a median ECS estimate of 1.6 oC, which makes the discrepancy with global climate models even larger. A recent update by Lewis and Curry (2018) with more recent data concluded that high estimates of ECS derived from a majority of global climate models are statistically inconsistent with observed warming during the historical period. Lewis and Curry further concluded that the observationally-constrained values of ECS imply 21st century warming under increased CO2 forcing of only 55-70% of the mean warming simulated by global climate models.
Apart from the uncertainties in the climate models described above, there are two overarching problems with these projections (Curry, 2018b):
- The scenarios of future climate are incomplete, focusing only on emissions.
- The ensemble of climate models do not sample the full range of possible values of ECS, neglecting values between 1 and 2.1 oC, with values between 1.5 and 2.1 oC being within the IPCC AR5 likely
The IPCC AR5 acknowledges the constraints, assumptions, contingencies and uncertainties of their projections of future climate change:
“With regard to solar forcing, the 1985–2005 solar cycle is repeated. Neither projections of future deviations from this solar cycle, nor future volcanic radiative forcing and their uncertainties are considered.”
“Any climate projection is subject to sampling uncertainties that arise because of internal variability. [P]rediction of the amplitude or phase of some mode of variability that may be important on long time scales is not addressed.”
The climate model projections of 21st century surface temperature and sea level rise are therefore contingent on the following assumptions [IPCC AR5 WG1 Section12.2.3]:
- Emissions follow the specified concentration pathways (RCP).
- Climate models accurately predict the amount of warming in 21st century.
- Solar variability follows that of the late 20th century, which coincided with a Grand Solar Maximum.
- Natural internal variability of ocean circulations does not impact temperature or sea level rise on these timescales.
- Major volcanic eruptions are not considered.
Each of these contingent assumptions, with the possible exception of natural internal variability, likely contributes to a warm bias in the 21st century climate model projections.
7.2 2100 – manmade climate change
As summarized in Section 4.5, our basic physical understanding of hurricane processes leads us to expect the following in a warmer climate:
- Potential Intensity theory leads to the expectation of increased intensity.
- Hurricanes are expected to produce more rainfall.
- No particular rationale for an increase or decrease in the number of hurricanes.
Quantitative projections of future changes in hurricane activity require:
- Projections of 21st century climate from both manmade and natural climate change
- An understanding of how and why hurricanes change with a changing climate.
As summarized in Chapter 4, our understanding of how and why hurricanes change in a changing climate is incomplete, with qualitative understanding based on analysis of limited observations and theoretical understanding. At best, climate model-based projections of future hurricane activity are contingent on the predicted amount of warming.
The IPCC AR5 provided a synthesis of global and regional model-based projections of future hurricane climatology by 2081 – 2100 relative to 2000 – 2019. Globally, their consensus projection is for decreases in hurricane numbers by approximately 5 – 30%, increases in the frequency of categories 4 and 5 storms by 0 – 25%, an increase of 0 – 5% in typical lifetime maximum intensity, and increases in rainfall rate by 5 – 20%.
Here are the conclusions from the IPCC AR5 (2013):
“Based on process understanding and agreement in 21st century projections, it is likely that the global frequency of occurrence of tropical cyclones will either decrease or remain essentially unchanged, concurrent with a likely increase in both global mean tropical cyclone maximum wind speed and precipitation rates. The future influence of climate change on tropical cyclones is likely to vary by region, but the specific characteristics of the changes are not yet well quantified and there is low confidence in region-specific projections of frequency and intensity.”
A summary of research since the IPCC AR5 is provided by the NCA4 (2017), whereby some studies have provided additional support for the AR5 conclusions, and some have challenged aspects of it. In the end, the NCA4 conclusions were identical to the IPCC AR5 conclusions cited above.
7.2.1 Hurricane formation and frequency
As summarized by Walsh et al. (2016), at present there is no climate theory that can predict the formation rate of tropical cyclones from the mean climate state. It has been known for many years that there are certain atmospheric conditions that either promote or inhibit the formation of tropical cyclones, but so far an ability to relate these quantitatively to mean rates of tropical cyclone formation has not been achieved, other than by statistical means through the use of empirically-based genesis potential indices (e.g. Menkes et al. 2012).
An important test of climate model predictions of future hurricane frequency is whether the climate models can simulate the present hurricane climatology. Simulation of the climatological number of Atlantic hurricanes is particularly difficult.
Most climate models predict a decrease in the global number of hurricanes by 2100. Explanations of this decrease are linked to reduced relative humidity in the mid-levels of the atmosphere and reduced upward rising motion in hurricane formation regions. Not all methods for determining hurricane numbers identify a decrease in future numbers, however. Emanuel (2013) uses a downscaling method in which incipient tropical vortices are “seeded” into large-scale climate conditions provided from a number of different climate models for current and future climate conditions. Emanuel’s approach generates more hurricanes in a warmer world when forced with the output of climate models.
While most models predict fewer tropical cyclones globally in a warmer world, the difference in the predictions among different climate models becomes more significant when smaller regions of the globe are considered. This appears to be a particular issue in the Atlantic basin, where climate model performance has been often poorer than in other oceanic regions. The issue as to whether the number of hurricanes will change in a warmer climate remains unresolved.
Using millennia-long climate model simulations, Lavender et al. (2018) examined whether the record number of tropical cyclones in the 2005 Atlantic season is close to the maximum possible number for the present climate of that basin. By estimating both the mean number of hurricanes and their possible year-to-year random variability, they found that the likelihood that the maximum number of storms in the Atlantic could be greater than the number of events observed during the 2005 season is less than 3.5%. Hence, the 2005 season can be used as a risk management benchmark for the maximum possible number of tropical cyclones in the Atlantic.
7.2.2 Hurricane intensity
GFDL (2018) provides an analysis of the predictions of hurricane changes by 2100:
“Hurricane intensities globally will likely increase on average by 1 to 10%, according to model projections for a 2 oC [4 oF] global warming. The global proportion of tropical cyclones that reach very intense (Category 4 and 5) levels will likely increase due to anthropogenic warming over the 21st century. There is less confidence in future projections of the global number of Category 4 and 5 storms, since most modeling studies project a decrease (or little change) in the global frequency of all tropical cyclones combined.”
With regards to the North Atlantic, GFDL (2018) provides the following assessment:
“Current climate models suggest that tropical Atlantic SSTs will warm dramatically during the 21st century, and that upper tropospheric temperatures will warm even more than SSTs. Furthermore, most of the climate models project increasing levels of vertical wind shear over parts of the western tropical Atlantic. Both the increased warming of the upper troposphere relative to the surface and the increased vertical wind shear are detrimental factors for hurricane development and intensification, while warmer SSTs favor development and intensification.”
“The GFDL hurricane model supports the notion of a substantial decrease (~25%) in the overall number of Atlantic hurricanes and tropical storms with projected 21st century climate warming. However, the hurricane model also projects that the lifetime maximum intensity of Atlantic hurricanes will increase by about 5% during the 21st century. At present we have only low confidence for an increase in category 4 and 5 storms in the Atlantic; confidence in an increase in category 4 and 5 storms is higher at the global scale.”
Using the GFDL hurricane modeling system, Knutson et al. (2015) found that projected median hurricane size is found to remain nearly constant globally, with increases in most basins offset by decreases in the northwest Pacific.
Changes in surface and subsurface ocean conditions can both influence a hurricane’s intensification. Huang et al. (2014) suggest a suppressive effect of subsurface oceans on the intensification of future hurricanes. Under global warming, the subsurface vertical temperature profile may contribute to a stronger ocean cooling effect during the intensification of future hurricanes. Emanuel (2015) estimated that the effect of such increased upper ocean stratification is relatively small, reducing the projected intensification of hurricanes by only about 10%–15%.
The largest increase in Category 4-5 Atlantic hurricanes is predicted by Bender et al. (2010). Owing to the large interannual to decadal variability of SST and hurricane activity in the basin, Bender et al. estimate that detection of an anthropogenic influence on intense hurricanes would not be expected for a number of decades, even assuming a large underlying increasing trend (+10% per decade).
An increase in rainfall from hurricanes in a warmer climate is a consistent finding from climate model simulations. As summarized by GFDL (2018), hurricane rainfall rates will likely increase in the future due to manmade global warming and the accompanying increase in atmospheric moisture content. Modeling studies on average project an increase on the order of 10-15% for rainfall rates averaged within about 100 km of the storm for a 2oC [4oF] global warming scenario.
7.3 2050 – decadal variability
Climate-model based projections of future hurricane activity have focused on the impacts of manmade climate change. It is of substantial interest to understand how hurricane activity might vary on timescales out to 2050, associated with the known modes of interannual and decadal variability in specific ocean basins.
The evolution of the climate on decadal time scales is the combined result of an externally forced component – due to greenhouse gases, aerosols and natural radiative forcing agents – and natural internal variability of the climate system. A decadal climate prediction attempts to simultaneously forecast the evolution of both of these components over the next few decades.
The Decadal Climate Prediction Project (DCPP) is a coordinated investigation into decadal climate prediction and variability. The first generation of the DCPP simulations was reported under CMIP5 in the IPCC AR5 [Chapter 11]. It was concluded that: “There is limited agreement and medium evidence that the Atlantic and Pacific patterns of climate variation exhibit predictability on timescales up to a decade.”
The next generation of the DCPP simulations is described by Boer et al. (2016), for the CMIP6 and the forthcoming IPCC AR6 Report. At this point, the climate models, even when the oceans are initialized with current observations, do not have any prediction skill beyond a decade at most. The biggest challenge is predicting shifts in the Atlantic and Pacific patterns of decadal variability (e.g. AMO, PDO).
7.3.1 Scenarios of modes of decadal variability
Given the challenges associated with climate model predictions on decadal scales, an alternative approach is to consider possible future scenarios of the indices of decadal climate variability and shifts in the multidecadal indices such as the AMO and PDO. Section 4.3 described the natural internal modes of variability, including the Atlantic Multidecadal Oscillation (AMO), North Atlantic Oscillation (NAO), Atlantic Multidecadal Mode (AMM), Pacific Decadal Oscillation (PDO) and the North Pacific Gyre Oscillation (NPGO).
Among these modes, a forthcoming shift in the phase of the AMO (away from the current warm phase to the cool phase; Figures 4.4, 4.5) would have the greatest impact on Atlantic hurricanes (Figure 4.6). Frajka-Williams et al. (2017) report a decline in the AMO index since 2013.
The timing of a shift to the AMO cold phase is not predictable; it depends to some extent on unpredictable weather variability. However, analysis of historical and paleoclimatic records suggest that a transition to the cold phase is expected prior to 2050. Enfield and Cid-Serrano (2006) used paleoclimate reconstructions of the AMO to develop a probabilistic projection of the next AMO shift. Figure 7.3 shows the probability of an AMO shift relative to the number of years since the last regime shift. The previous regime shift occurred in 1995; hence in 2019, it has been 24 years since the previous shift. Figure 7.3 indicates that a shift to the cold phase should occur within the next 15 years, with a 50% probability of the shift occurring in the next 7 years.
The implications of a shift to the cool phase of the AMO on Atlantic hurricanes include:
- Fewer major hurricanes and lower values of Accumulated Cyclone Energy (ACE) (Figure 4.6)
- Fewer landfalls striking Florida, the U.S. east coast and the Caribbean (Figures 5.6, 5.8)
Figure 7.3 Probability of an AMO regime shift relative the number of years since the last regime shift. Source: Enfield and Cid-Serrano (2006)
7.3.2 Scenarios of interannual variability
Atlantic hurricane statistics for the period to 2050 depend not only on the timing of a shift of the AMO to the cool phase, but also on the variability of the other climate indices.
Caron et al. (2014) found that while some influences, such as ENSO, remain present regardless of the AMO phase, other climate factors show an influence during only one of the two phases. During the negative phase, Sahel precipitation and the NAO play a role, while during the positive phase, the 11-year solar cycle and dust concentration over the Atlantic appear to be more important.
Lim et al. (2016) showed that the NAO and AMM can strongly modify and even oppose the well-known ENSO impacts. While the predictability of these modes is limited to seasons rather than years, the statistics of combinations of these indices provides useful information regarding the interannual variability within the decadal time horizon. Of particular interest is the frequency of occurrence of years of extremely high or low hurricane activity. Patricola et al. (2014) investigated the possible effects of combinations of extreme phases of the AMM and ENSO. Individually, the negative AMM phase and El Nino each inhibit Atlantic hurricanes, and vice versa. Simultaneous strong El Nino and strongly positive AMM, as well as strong concurrent La Nina and negative AMM, produce near-average Atlantic ACE, suggesting compensation between the two influences. Strong La Nina and strongly positive AMM together produce extremely intense Atlantic hurricane activity, while strong El Nino and negative AMM together are not necessary conditions for significantly reduced Atlantic tropical cyclone activity.
The past decade or so has seen a preponderance of El Nino events (relative to La Nina). The PDO has been weakly negative for the past year, following a period since 2014 of mostly positive values. Presumably, at some point in the next 30 years, we can expect a period when La Nina events dominate.
The general probabilistic approach used by Enfield and Cid-Serrano (2006) seems promising for developing probabilities of regime combinations, which can then be related to Atlantic hurricane activity via historical relationships with these regime indices. However, the possibility of data-driven climate dynamics-based probabilistic predictions and scenarios of decadal scale hurricane activity is largely untapped.
7.4 Landfall impacts
The most unambiguous signal for hurricane landfall impacts in a warmer climate is that projected sea level rise should be causing higher storm surge levels for hurricanes that do occur, all else being equal. As summarized by Curry (2018; Section 5.7):
“Emissions scenario choice exerts a great deal of influence on predicted sea level rise after 2050. If RCP8.5 is rejected as an extremely unlikely scenario, then the appropriate range of sea level rise scenarios to consider for 2100 is 0.2–1.6 m [8 inches to 5 feet]; however, values exceeding 2 feet are increasingly weakly justified. Values exceeding 5 feet require a cascade of poorly understood and extremely unlikely to impossible events. Further, these values of sea level rise are contingent on the climate models predicting the correct amount of temperature increase.”
Increased rainfall rates can also be expected in a warmer climate (Section 7.2.3). There is no evidence of increasing hurricane size, which influences storm surge, rainfall amounts and the number of tornadoes (Section 4.5).
If climate model projections of fewer hurricanes but a greater percentage of Category 4 and 5 storms are correct, the tradeoff between these two competing effects on overall landfall impacts is not straightforward. The statistics of rare Category 4 and 5 landfalling events are much more volatile than basin-wide hurricane metrics.
Emanuel (2011) estimated the time of emergence of global warming effects on U.S. hurricane damage. Using a recently developed hurricane synthesizer driven by outputs from global climate models, 1000 artificial 100-yr time series of Atlantic hurricanes that make landfall along the U.S. Gulf and East Coasts were generated for four climate models and for current climate conditions as well as for the warmer climate circa 2100. These synthetic hurricanes produce damage to a portfolio of insured property according to an aggregate wind-damage function; damage from flooding was not considered. Three of the four climate models used produced increasing damage with time, with the global warming signal emerging on time scales of 40, 113, and 170 years. For the fourth climate model, damages decreased with time, but the signal was weak.
Substantial advances have been made in recent years in the ability of climate models to simulate the variability of hurricanes. However, inconsistent hurricane projections emerge from modeling studies due to different down-scaling methodologies and warming scenarios, inconsistencies in projected changes of large-scale conditions, and differences in model physics and tracking algorithms. Systematic numerical modeling experiments organized under the auspices of the Hurricane Working Group of the U.S. CLIVAR Program (Walsh et al. 2015) were designed to coordinate efforts to produce a set of model experiments designed to improve understanding of the variability of tropical cyclone formation in climate models. Progress continues to be made, particularly with models that are coupled to the ocean.
Apart from the challenges of simulating hurricanes in climate models, the amount of warming projected for the 21st century is associated with deep uncertainty. Hence, any projection of future hurricane activity is contingent on the amount of predicted global warming being correct.
Recent assessment reports have concluded that there is low confidence in future changes to hurricane activity, with the greatest confidence associated with an increase in hurricane-induced rainfall and sea level rise that will impact the magnitude of future storm surges. Any projected change in hurricane activity is expected to be small relative to the magnitude of interannual and decadal variability in hurricane activity, and is at least several decades away from being detected.
Decadal variability of hurricane activity over the next several decades could provide much greater variability than the signal from global warming over the next century. In particular, a shift to the cold phase of Atlantic Multidecadal Oscillation (AMO) is anticipated within the next 15 years. All other things being equal (such as the frequency of El Nino and La Nina events), the cold phase of AMO harkens reduced Atlantic hurricane activity and fewer landfalls for Florida, the east coast and the Caribbean.