Will a Warmer World Be Stormier?

delgenioArticles, Climate, Extreme Weather Theme, Original

Picture of lightning

Anthony D. Del Genio
NASA Goddard Institute for Space Studies
New York, NY

Picture of lightning

Fig. 1. Lightning storms occur when conditions near the surface are so warm and humid that air rises fast enough to carry water droplets up to altitudes at which ice crystals have formed. Photo courtesy NOAA/National Weather Service.


Increases in carbon dioxide and other greenhouse gases have almost certainly played a major role in the observed temperature increases of the 20th Century, according to the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report. Global climate model (GCM) projections suggest that continued 21st Century increases in greenhouse gases will further warm the climate by a few degrees. A temperature change this small may not seem very serious, since local weather can fluctuate by much more than this from day to day. The small global mean change, however, is expected to create large problems in sensitive areas of the Earth system — rising sea level leading to increased coastal flooding, more heat waves and drought, and the disappearance of summer Arctic sea ice, to name a few.

Among the most important potential impacts of climate warming are changes in extreme weather events. Global precipitation will increase, and the heaviest precipitation events are intensifying [1], but with regional differences: Wet regions such as the tropical rainforests will become rainier while semi-arid regions of the subtropics expand and become drier. This is only one of the impacts of extreme weather, however. Equally important are the winds, hail, lightning, and fires that result from storms. Will we see more frequent storms in a warmer climate? Stronger storms? Will storm damage escalate? These questions are much harder to answer. Different types of storms — thunderstorms, tornadoes, hurricanes, synoptic midlatitude storms — occur under different conditions,

Map of the earth showing lightning flashes per square kilometer

Fig. 2. Lightning flashes per square kilometer per year for 1995-2002 observed by NASA satellites. Most lightning occurs over land, even though more rain falls over the oceans. Courtesy NASA Earth Observatory.

though all storms can be thought of as the atmosphere’s most efficient means of redistributing heat from places with an excess to places with a deficit. Global climate models only explicitly represent the atmosphere on spatial scales of 100-200 kilometers, due to the computational demands of conducting many long simulations of future climate under various emission scenarios. By comparison, most weather is generated by processes that occur over kilometers to tens of kilometers. Thus, changes in weather can only be indirectly inferred in climate models from changes in large-scale environmental conditions.

Finally, unlike precipitation, for which long and reliable historical records exist in some parts of the world, records for other aspects of weather are too short to detect trends or contain observational biases that render trends meaningless.

Extreme weather by its very definition is rare but occurs regardless of any climate change. Thus, no single event – a Hurricane Katrina, a January tornado in Wisconsin, a record East Coast blizzard – can be interpreted as evidence of climate change. Climate scientists, not endowed with the powers of a Ghost of Climate Yet to Come, sometimes use such events as icons to raise awareness of what a warmer world may look like to the climate equivalents of Ebenezer Scrooge in today’s world. Whether this does more good than harm can be debated, but it is not our basis for predicting what might occur in the future.

Graph showing Number of occurrences of different freezing level updraft speeds in convective storms simulated by the Goddard Institute for Space Studies global climate model

Fig. 3. Number of occurrences of different freezing level updraft speeds in convective storms simulated by the Goddard Institute for Space Studies global climate model for current (solid) and doubled CO2 (dashed) conditions (5). Courtesy American Geophysical Union.

Despite the obstacles, considerable progress has been made in understanding how future climate change may affect weather. In this article, we discuss what is known and not known about how climate change will alter extreme weather events beyond just the precipitation that they produce.

Thunderstorms, lightning, fires
Thunderstorms (Fig. 1) receive less attention than other types of extreme weather because they are so common and the damage they cause so localized. Perhaps because they are taken for granted, lightning strikes are the second leading cause of weather-related deaths in the United States (after floods) [2]. U.S. lightning mortality has decreased by almost a factor of 3 over the past 3 decades, but probably due to better forecasts and societal/demographic factors rather than because of any change in lightning occurrence to date [3].

Most torrential short-lived rainstorms do not produce thunder and lightning. Data from NASA’s Tropical Rainfall Measuring Mission (TRMM) show that annual mean rainfall is greater over the oceans than over land. Lightning, however, is an order of magnitude more likely to occur in continental rainstorms than maritime storms (Fig. 2). These rainstorms are a manifestation of the process of convection in the atmosphere. Heating of Earth’s surface by sunlight and thermal infrared radiation must be offset by energy losses to maintain a stable temperature. The surface does this partly by evaporating water, which then condenses as buoyant air rises. This removes excess heat from the surface to higher altitudes. The rising air carries water droplets upward until they are too large to stay aloft and fall out as precipitation. When updrafts are vigorous, water drops are easily carried above the freezing level and some collide with fluffy ice crystals to make larger, denser ice pellets called graupel. The presence of both graupel and fluffy ice crystals that fall at different speeds appears to be a necessary ingredient for lightning.

Picture of a mature Texas tornado

Fig. 4. Mature Texas tornado. Tornadoes and other severe thunderstorms occur when warm, humid air near the ground and cold air aloft combine with winds that change speed and direction with height. Photo courtesy NOAA National Severe Storms Laboratory.

The land-ocean difference in lightning helps us anticipate how lightning might change as the climate warms. The massive ocean stores heat and barely warms over the course of the day. Land surfaces, which cannot store heat, warm significantly from morning to afternoon, and the air that then rises in convective storms is warmer and more buoyant than that over oceans, creating strong updrafts more prone to generate lightning. As CO2 increases and more heat is radiated down to the surface, the ocean will evaporate more water, while the drier land surface cannot do that to the same extent and must warm more as a result [4]. This should give rise to stronger convective updrafts and more lightning. Although small-scale convective updraft speeds cannot be directly simulated by GCMs, models can diagnose them from large-scale atmospheric temperature and humidity profiles. In a doubled CO2 climate in one such model, updraft speeds over land are indeed slightly faster [5]. In the western U.S., the model predicts fewer storms with strong updrafts overall, but about 25% more of the strongest storms (Fig. 3). Since lightning increases nonlinearly with updraft strength, the net result is about a 5% increase in lightning flash rate.

table showing Simulated changes in factors affecting severe thunderstorm occurrence in the United States between the periods 1962-1989 and 2072-2099 for northern spring (left panels) and northern summer (right panels) conditions.

Fig. 5. Simulated changes in factors affecting severe thunderstorm occurrence in the United States between the periods 1962-1989 and 2072-2099 for northern spring (left panels) and northern summer (right panels) conditions. First row: Convective available potential energy, a measure of how buoyant air lifted from the surface would become, aggregated over the depth over which it remains buoyant. Second row: Concentration of water vapor at the ground in grams of water per kilogram of air. Third row: Wind shear, defined as the magnitude of the vector difference in wind between the surface and 6 kilometers altitude. Fourth row: Resulting number of days with environments conducive to severe thunderstorms (12). Courtesy National Academy of Sciences.

Regardless of the extent to which lightning increases with warming, lightning damage should increase because of its role in igniting forest fires. In western North America, wildfire area burned has increased dramatically in recent decades [6,7]. The primary reason is drying of the surface as temperature rises, creating more “fuel” for such fires. Some of the recent drying can be attributed to natural variability. But climate models show a consistent tendency for areas such as the southwest U.S. to dry out in the 21st Century as the sinking branch of the tropical Hadley circulation expands poleward and carries drier air down to the surface [8].

A recent GCM study projects tens to hundreds of additional fire counts per year, per 4°x5° latitude-longitude area, in the western North America and much of Eurasia, by the end of the century [9]. The eastern U.S. is predicted to become more humid and rainier instead, resulting in fewer fires. Only 12% of U.S. wildfires are ignited by natural causes, but these account for 52% of the acres burned [10], so even a small climate change in lightning flash rate has important consequences.
Severe thunderstorms and tornadoes

Thunderstorms sometimes occur in the presence of strong wind shear, i.e., a change in wind speed and/or direction with height. Wind shear creates a horizontal “tube” of rotating air. When a thunderstorm arises in these circumstances, the rising air in the storm tilts the rotating tube of air into the vertical. The result can be a “significant severe thunderstorm” with at least one of the following characteristics: Hail greater than 5 cm in diameter, wind gusts exceeding 120 km/hr, or a tornado (Fig. 4) of strength F2 or greater on the Fujita scale. Severe thunderstorms are localized and difficult to predict, but the climatological, geographical and seasonal distribution of environments conducive to significant severe thunderstorms can be inferred by combining large-scale estimates of wind shear and how unstable the atmosphere is to thunderstorms [11].

As noted earlier, increasing CO2 concentrations should make thunderstorms more vigorous. Wind shear, however, is controlled by the rotation of the Earth and by horizontal temperature differences, mostly between low and high latitudes. As the climate changes, warming will not be uniform. In the tropics, convection that removes surface heat to high altitudes acts as a thermostat, limiting surface warming. In the polar regions, where few such storms occur, heating by greenhouse gases remains at the surface, and is exacerbated by the melting of bright sea ice that exposes more of the dark ocean surface and causes more sunlight to be absorbed. The net result is that in the lower troposphere, the temperature difference between low and high latitudes decreases as the planet warms, creating less wind shear.

Figure showing Recent trends in annual tornado reports, annual number of days with tornadoes, annual reports of tornadoes of severity F2 or greater, and population for the Southern Great Plains region (upper panel) and the entire United States (lower panel)

Fig. 6. Recent trends in annual tornado reports, annual number of days with tornadoes, annual reports of tornadoes of severity F2 or greater, and population for the Southern Great Plains region (upper panel) and the entire United States (lower panel) (13). Courtesy American Geophysical Union.

We thus have two competing effects on the climate change in significant severe thunderstorms – which one wins? A study using a regional climate model and the same indices of significant severe thunderstorms as have been used to diagnose their patterns in the current climate finds that, especially in the central and eastern United States (the preferred locations for severe weather anywhere in the world), we can expect a few more days per month with conditions favorable to severe thunderstorm occurrence in a doubled CO2 climate [12] (Fig. 5).

A more conservative estimate that just accounts for climate changes in instability in the lower to middle troposphere finds that on average, although thunderstorms will have stronger updrafts, wind shear that is more often below the threshold for severe storm behavior will dominate [5]. However, combined occurrences of the very strongest wind shears and updraft speeds will increase with warming, suggesting more of the most significant severe thunderstorms.

At the end of severe thunderstorm spectrum are tornadoes, which though much rarer than thunderstorms, account for almost as many fatalities and much more damage [2]. Tornadoes are more difficult to predict than other severe thunderstorms, even climatologically [11]. But does the past provide any hints? A record spanning the past half-century shows a dramatic increase (about 14 per year) in U.S. tornado reports during this time [13] (Fig. 6). Curiously, though, there has been no trend in the number of days with reported tornadoes and even a small decrease in the number of reported strong (category F2-F5) tornadoes. A possible clue to the apparent discrepancy is that the increase in overall tornado reports roughly matches that of the U.S. population over this time, suggesting that the trend may be an artifact of greater tornado detection due to increases in population density, awareness of severe weather threats, and modern technological advances such as Doppler radar. At the current time, it is therefore not possible to anticipate even the sign of any climate change in tornado occurrence or strength.

Larger-scale storms

Hurricanes (Fig. 7) and other tropical cyclones can be thought of as heat engines that take energy in by evaporating warm ocean water, and eject it at a colder temperature near the tropopause after air rises and water condenses in the eyewall [14]. The long-term record of Atlantic tropical cyclones is observationally biased, because those that never made landfall were often not detected before the start of the satellite era [15]. Nonetheless, an upward trend in Atlantic and West Pacific tropical cyclone power dissipation in the past few decades — based on the frequency, duration, and intensity of observed storms — is well correlated with increases in sea surface temperature [16]. Extrapolating these relationships to a warmer climate using projections of rising ocean temperatures from global climate models suggests a dramatic increase in hurricane destructiveness over the 21st Century.

MODIS satellite image of Hurricane Katrina shortly before landfall in New Orleans.

Fig. 7. MODIS satellite image of Hurricane Katrina shortly before landfall in New Orleans. Courtesy NASA Earth Observatory.

Recent research suggests, however, that this may be misleading. The cold upper troposphere outflow temperature is as important to a tropical cyclone heat engine as the warm surface input temperature. The outflow temperature does not depend just on the local state but rather on conditions throughout the tropics, since the tropical general circulation redistributes heat efficiently at high altitudes. The relevant factor may therefore not be how warm the water is in a given ocean basin, but how warm it is relative to other ocean basins. In the late 20th Century, the Atlantic was anomalously warm, but this is not expected to continue in the long-term. Projections emphasizing relative rather than absolute sea surface temperature changes suggest little change in hurricane destructiveness in the 21st Century [17] (Fig. 8).

These inferences are not based on actual simulations of tropical cyclones, whose core regions of strong winds cannot be resolved by today’s global climate models. One approach to this problem is “downscaling,” a procedure in which climate changes in large-scale atmosphere and ocean conditions predicted by a global model are used as input to a fine-scale regional model that does resolve tropical cyclones. One such study applied to several of the IPCC models concludes that there will be fewer Atlantic tropical cyclones overall in a warmer climate, but more of the strongest (category 4-5) hurricanes [18]. As a caution, one of the four models examined (a respected model) predicts fewer strong hurricanes in a warmer world instead.

Midlatitude synoptic storms (Fig. 9), associated with low and high pressure centers and warm and cold fronts, are the largest in scale (a thousand kilometers or more) of all storm types. Ostensibly these storms are resolved explicitly by the IPCC models, but much of their precipitation is generated in the frontal regions, which are only tens of kilometers in scale.

Graph showing the departure of the observed Atlantic tropical cyclone power dissipation index from mean conditions for the period 1946-2007

Fig. 8. The black curve in both panels shows the departure of the observed Atlantic tropical cyclone power dissipation index from mean conditions for the period 1946-2007. The upper and lower panels show two different statistical relationships between the power dissipation index and sea surface temperature, one based on actual sea surface temperature (red curve, upper panel) and another based on Atlantic sea surface temperature relative to tropical mean sea surface temperature (cyan curve, lower panel). The orange and blue curves represent global climate model inferences of the power dissipation index based on the same statistical relationships (thin curves for individual models, thick curves for the mean of all models). Independent estimates of future Atlantic tropical cyclone power dissipation index using a different technique are shown in green at the right (17). Courtesy American Association for the Advancement of Science.

Synoptic storms arise from potential energy contained in horizontal temperature differences between warm and cold places (south vs. north, or ocean vs. land). Warm air displaced poleward into colder air ahead of a low rises, and cold air displaced equatorward behind a low into warmer air sinks, converting potential into kinetic energy and effecting a net poleward transport of heat.
The conventional wisdom is that since the polar regions should warm more than the tropics in the coming century, synoptic storms should weaken. There are several complicating factors, though.
First, equator-pole temperature differences weaken only near the surface. At higher altitudes, the reverse actually occurs, and in one IPCC model this causes the mid-latitude storm tracks to intensify with warming [19]. Second, since a warmer atmosphere will contain more water vapor, the latent heat released when the added water condenses and precipitates in rising air along warm fronts may intensify future synoptic storms. Most of the IPCC AR4 models simulate fewer synoptic storms overall, but more of the strongest storms, as the climate warms. A recent study using a higher-resolution model, however, finds no evidence of future intensification except for precipitation [20].

One topic of considerable recent interest is whether the strong snowstorms of the past two winters along the east coast of the U.S. are related to climate change. These events seem to have become a Rorschach test of underlying opinions about the reality of global warming. More precipitation is expected in a warmer climate, but it is less clear whether that implies more snow (because of more precipitation in general) or less snow (because winter storms will more often occur above the freezing temperature and thus produce rain). The question of the recent winters is more straightforward, though, because no short time period of unusual weather can ever be interpreted in terms of climate change. Winter weather in eastern North America and Europe is affected by the North Atlantic Oscillation (NAO), a seesaw in pressure between the climatological Icelandic low and Azores high. When the NAO is in its negative phase, the low and high are weaker than normal, and cold air outbreaks and snow are more likely in the eastern U.S. This has been shown to be responsible for the snowy 2009-2010 winter [21] and may also explain the equally severe 2010-2011 winter along the U.S. eastern seaboard. In many other parts of the world, these winters were warmer than normal. The NAO is relatively unpredictable and it is difficult to project whether it will systematically change in a warming world.


GOES satellite image of the December 27, 2010 blizzard along the east coast of the United States.

Fig. 9. GOES satellite image of the December 27, 2010 blizzard along the east coast of the United States. Courtesy NASA Goddard Space Flight Center.

Despite their diverse spatial scales, wind patterns, and energy sources, a common narrative thread has emerged for all types of storms: As the climate warms, we might experience fewer storms overall, but more of the strongest storms. Put another way, a warmer climate will place greater demands on the atmosphere to transport heat upward and poleward, but this will be done more efficiently, in a smaller number of events that each accomplish more of the required transport. This conclusion is tentative — the science of prediction of climate changes in storms is in its infancy. Not every model agrees with the “consensus” result, and until and unless the conceptual basis for a shift to fewer but stronger storms is firmly established, the projections of global climate models should be regarded as educated guesses.

What the public cares about most is the damage caused by storms. This depends on more than just changes in the storms themselves, and here we may be on firmer footing. Regardless of how lightning changes with warming, drying in forested areas of western North America will likely lead to more fire damage [8,9]. Regardless of whether hurricanes and synoptic storms intensify with warming, sea level rise [22] and increased population and development [23] imply more flooding damage to coastal areas from storm surges. A rational public and private sector response to the threat of storm damage in a changing climate must therefore acknowledge scientific uncertainties that are likely to persist beyond the time at which decisions will need to be made, focus more on the risks and benefits of planning for the worst case scenarios, and recognize that the combination of societal trends and the most confident aspects of climate change predictions makes future economic impacts substantially more likely than does either one alone.

[1] S.-K. Min, X. Zhang, F.W. Zwiers, and G.C. Hegerl, “Human contribution to more-intense precipitation extremes,” Nature, vol. 470, pp. 378-381, Feb. 2011.
[2] E.B. Curran, R.L. Holle, and R.E. López, “Lightning casualties and damages in the United States from 1959 to 1994,” J. Climate, vol. 13, pp. 3448-3464, Oct. 2000.
[3] W.S. Ashley and C.W. Gilson, “A reassessment of U.S. lightning mortality,” Bull. Amer. Meteor. Soc., vol. 90, pp. 1501-1518, Oct. 2009.
[4] J.T. Fasullo, “Robust land-ocean contrasts in energy and water cycle feedbacks,” J. Climate, vol. 23, pp. 4677-4693, Sept. 2010.
[5] A.D. Del Genio, M.-S. Yao, and J. Jonas, “Will moist convection be stronger in a warmer climate?,” Geophys. Res. Letters, vol. 34, L16703, doi:10.1029/2007GL030525, 2007.
[6] N.P. Gillett, A.J. Weaver, F.W. Zwiers, and M.D. Flannigan, “Detecting the effect of climate change on Canadian forest fires,” Geophys. Res. Letters, vol. 31, L18211, doi:10.1029/2004GL020876, 2004.
[7] A.L. Westerling, H.G. Hidalgo, D.R. Cayan, and T.W. Sweetnam, “Warming and earlier spring increase western U.S. forest wildfire activity,” Science, vol. 313, pp. 940-943, Aug. 2006.
[8] R. Seager et al., “Model projections of an imminent transition to a more arid climate in southwestern North America,” Science, vol. 316, pp. 1181-1184, May 2007.
[9] O. Pechony and D.T. Shindell, “Driving forces of global wildfires over the past millennium and the forthcoming century,” Proc. Natl. Acad. Sci., vol. 107, pp. 19167-19170, Nov. 2010.
[10] U.S. Fire Administration, “2000 wildland fire season,” U.S. Fire Administration Topical Fire Research Series, vol. 1, Issue 2, 4 pp.
[11] H.E. Brooks, J.W. Lee, and J.P. Craven, “The spatial distribution of severe thunderstorm and tornado environments from global reanalysis data,” Atmos. Res., vol. 67/68, pp. 73-94, 2003.
[12] R.J. Trapp, N.S. Diffenbaugh, H.E. Brooks, M.E. Baldwin, E.D. Robinson, and J.S. Pal, “Changes in severe thunderstorm environment frequency during the 21st century caused by anthropogenically enhanced global radiative forcing,” Proc. Natl. Acad. Sci., vol. 104, pp. 19719-19723, Dec. 2007.
[13] N.S. Diffenbaugh, R.J. Trapp, and H.E. Brooks, “Does global warming influence tornado activity?,” EOS, vol. 89, pp. 553-554, Dec. 2008.
[14] K. Emanuel, “Hurricanes: Tempest in a greenhouse,” Phys. Today, vol. 59, pp. 74-75, Aug. 2006.
[15] C.W. Landsea, “Counting Atlantic tropical cyclones back to 1900,” EOS, vol. 88, pp. 197-208, May 2007.
[16] K. Emanuel, “Factors affecting tropical cyclone power dissipation,” J. Climate, vol. 20, pp. 5497-5509, Nov. 2007.
[17] G.A. Vecchi, K.L. Swanson, and B.J. Soden, “Whither hurricane activity?,” Science, vol. 322, pp. 687-689, Oct. 2008.
[18] M.A. Bender, T.R. Knutson, R.E. Tuleya, J.J. Sirutis, G.A. Vecchi. S.T. Garner, and I.M. Held, “Modeled impact of anthropogenic warming on the frequency of intense Atlantic hurricanes,” Science, vol. 327, pp. 454-458, Jan. 2010.
[19] Y. Wu, M. Ting, R. Seager, H.-P. Huang, and M.A. Cane, “Changes in storm tracks and energy transports in a warmer climate simulated by the GFDL CM2.1 model,” Clim. Dyn., doi:10.1007/s00382-010-0776-4, March 2010.
[20] L. Bengtsson, K.I. Hodges, and N. Keenlyside, “Will extratropical storms intensify in a warmer climate?,” J. Climate, vol. 22, pp. 2276-2301, May 2009.
[21] R. Seager, Y. Kushnir, J. Nakamura, M. Ting, and N. Naik, “Northern Hemisphere winter snow anomlies: ENSO, NAO, and the winter of 2009/10,” Geophys. Res. Letters, vol. 37, L14703, doi:10.1029/2010GL043830, 2010.
[22] S. Rahmstorf, “A new view on sea level rise,” Nature Reports Climate Change, vol. 4, pp. 44-45, April 2010.
[23] R.A. Pielke Jr., J. Gratz, C.W. Landsea, D. Collins, M.A. Saunders, and R. Musulin, “Normalized hurricane damage in the United States: 1900-2005,” Natural Hazards Review, vol. 9, pp. 29-42, Feb. 2008.
Anthony Del Genio is a Research Physical Scientist at the NASA Goddard Institute for Space Studies (GISS). His research focuses on the representation of clouds and convective storms in the GISS global climate model and their role in cloud feedbacks on climate change, as well as climate impacts associated with storms. Del Genio is Principal Investigator on the NASA Tropical Rainfall Measuring Mission and Global Precipitation Mission, and on the NASA CloudSat/CALIPSO Mission. He also is Co-Chair of the Cloud Lifecycle Working Group of the DOE Atmospheric Systems Research Program. In addition, he studies dynamics and storms on other planets as an imaging team member on the NASA Cassini Orbiter Mission to Saturn and Titan. Del Genio has served as an Editor for the Journal of Climate and is the author of 115 peer-reviewed articles. He was elected a Fellow of the American Meteorological Society in 2007 and received the NASA Medal of Exceptional Scientific Achievement in 2008.