Volume 123, Issue 18 p. 10,314-10,325
Research Article
Free Access

Did Smoke From City Fires in World War II Cause Global Cooling?

Alan Robock

Corresponding Author

Alan Robock

Department of Environmental Sciences, Rutgers University, New Brunswick, NJ, USA

Correspondence to: A. Robock,

[email protected]

Contribution: Conceptualization, Methodology, ​Investigation, Resources, Writing - original draft, Writing - review & editing, Visualization, Supervision, Funding acquisition

Search for more papers by this author
Brian Zambri

Brian Zambri

Department of Environmental Sciences, Rutgers University, New Brunswick, NJ, USA

Now at Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA, USA

Contribution: ​Investigation, Resources, Writing - review & editing, Visualization

Search for more papers by this author
First published: 31 August 2018
Citations: 3

Abstract

Between 3 February and 9 August 1945, an area of 461 km2 in 69 Japanese cities, including Hiroshima and Nagasaki, was burned during the U.S. B-29 Superfortress air raids. In the previous 5 years, 205 km2 in German cities were destroyed, so the smoke that was generated was spread out over a much longer period of time than that from Japan in 1945. Observations of solar irradiance show reductions consistent with the hypothesis that smoke was injected into the stratosphere by the city fires. Historical simulations from the Coupled Model Intercomparison Project 5, with no smoke in their forcing, showed no postwar cooling. Global average surface air temperature observations during and following World War II are problematic, because of issues with measuring sea surface temperatures, but there were no large volcanic eruptions, El Niño, or La Niña during this period to confuse the record. Nevertheless, 1945 and 1946 global average land surface air temperatures were not significantly lower than the average for 1940–1944. Estimates of the amount of smoke generated by the fires are somewhat uncertain. Although the climate record is consistent with an expected 0.1–0.2 K cooling, because of multiple uncertainties in smoke injected to the stratosphere, solar radiation observations, and surface temperature observations, it is not possible to formally detect a cooling signal from World War II smoke.

Key Points

  • Between 3 February and 9 August 1945, an area of 461 km2 in 69 Japanese cities, including Hiroshima and Nagasaki, was burned during the U.S. B-29 Superfortress air raids, producing massive amounts of smoke
  • Because of multiple uncertainties in smoke injected to the stratosphere, solar radiation observations, and surface temperature observations, it is not possible to formally detect a cooling signal from World War II smoke
  • These results do not invalidate nuclear winter theory that much more massive smoke emissions from nuclear war would cause large climate change and impacts on agriculture

Plain Language Summary

Nuclear winter theory says that smoke from burning cities targeted with nuclear weapons would rise into the upper atmosphere and spread around the world, absorbing sunlight and cooling the surface. Unfortunately, we have two real-world examples of this, Hiroshima and Nagasaki, which were burned by U.S. atomic bombs on 6 and 9 August 1945. We discovered that this was actually the culmination of a genocidal U.S. bombing campaign. Between 3 February and 9 August 1945, an area of 461 square kilometers in 69 Japanese cities, including Hiroshima and Nagasaki, was burned during the U.S. B-29 Superfortress air raids, and the Hiroshima and Nagasaki smoke was less than 5% of the total. We calculated how much smoke was emitted and how much climate change would have been expected. Although there was a small cooling, because of multiple uncertainties in smoke injected to the stratosphere, sunlight observations, and surface temperature observations, we found that it is not possible to say for sure that this was a cooling signal from World War II smoke. However, these results do not invalidate nuclear winter theory that much more massive smoke emissions from nuclear war would cause large climate change and impacts on agriculture.

1 Introduction

Crutzen and Birks (1982) suggested that a nuclear war would start massive forest fires, creating toxic air and dark smoke clouds over much of the war zone. Turco et al. (1983) recognized that burning cities would produce even more soot than burning forests and that the soot would rise into the stratosphere where it would cover the Earth and produce global climate change so large that the climatic consequences were described as “nuclear winter.” Aleksandrov and Stenchikov (1983) conducted the first three-dimensional global climate modeling and found that large climate changes over the land were not moderated by the oceans. This new research confronted the world with the prospect of potential indirect effects of nuclear war much larger than the direct effects. While the direct effects might kill hundreds of millions in combat zones, the indirect effects would lead to collapse of world agriculture and starvation of billions of people even in regions that were not involved directly in the war. By 1990 the arms race and Cold War ended. Since then, the global nuclear arsenal has been reduced by a factor of 4 (e.g., Toon et al., 2008).

The world currently possesses about 14,000 nuclear weapons, distributed among nine nations (http://www.ploughshares.org/world-nuclear-stockpile-report, accessed 23 April 2018). When exploded on cities and industrial areas, the acknowledged targets of these nations, they would start fires, producing massive amounts of smoke. That smoke would block out sunlight, making it cold and dark at the surface for many years, as well as destroy ozone, enhancing ultraviolet radiation reaching the surface (e.g., Mills et al., 2014). The size of these impacts would depend on the number and yield of the nuclear weapons used, as well as the specific targets.

In the past decade, climate model simulations, which used a coupled atmosphere-ocean general circulation model run continuously for multiple 10-year simulations and with a model top at the mesopause, have supported the conclusion that full-scale global nuclear war can still produce a nuclear winter (Robock, Oman, & Stenchikov, 2007). In addition, several groups found that a nuclear war between new nuclear states such as India and Pakistan, using much less than 1% of the current global nuclear arsenal, could produce climate change unprecedented in recorded human history (Mills et al., 2014; Pausata et al., 2016; Robock, Oman, Stenchikov, et al., 2007; Stenke et al., 2013; Toon, Robock, et al., 2007; Toon, Turco, et al., 2007), global-scale ozone depletion (Mills et al., 2008, 2014), and widespread famine (Özdoğan et al., 2013; Xia et al., 2015; Xia & Robock, 2013).

All these results are model simulations. Is there any way to test the climatic response to stratospheric soot from observations? Unfortunately, history presents us with examples of cities that have burned in the past. Accidental fires burned numerous cities, including London in 1666, Chicago in 1871, and San Francisco in 1906. About the San Francisco fire, which was started by an earthquake, London (1906) wrote, “Not in history has a modern imperial city been so completely destroyed. San Francisco is gone.” Yet, during World War II, cities were burned intentionally (Caidin, 1960; The U.S. Strategic Bombing Survey, 1946).

“In the attacks on German cities incendiary bombs, ton for ton, were found to have been between four and five times as destructive as high explosive. … in the more serious fire raids, any fire-fighting equipment was found to have been of little avail. Fire storms occurred, the widespread fires generating a violent hurricane-like draft, which fed other fires and made all attempts at control hopeless. … 485,000 residential buildings were totally destroyed by air attack and 415,000 were heavily damaged, making a total of 20 percent of all dwelling units in Germany. In some 50 cities that were primary targets of air attack, the proportion of destroyed or heavily damaged dwelling units is about 40 percent. The result of all these attacks was to render homeless some 7,500,000 German civilians.” (The U.S. Strategic Bombing Survey, 1945).

And certainly these fires produced smoke plumes. In Japan, “Greatest source of alarm to our flyers were the terrific thermals, or hot-air currents, that rose from the blazing targets and sent our aircraft into a black hell of smoke” (Caidin, 1960, p. 154). “We headed into a great mushroom of boiling, oily smoke, and in a few seconds were tossed 5,000 feet [1,500 m] into the air” (Caidin, 1960, pp. 154–155). As London (1906) described for the San Francisco fire,

“Within an hour after the earthquake shock the smoke of San Francisco's burning was a lurid tower visible a hundred miles [160 km] away. And for three days and nights this lurid tower swayed in the sky, reddening the sun, darkening the day, and filling the land with smoke. … I watched the vast conflagration from out on the bay. It was dead calm. Not a flicker of wind stirred. Yet from every side wind was pouring in upon the doomed city. East, west, north, and south, strong winds were blowing upon the doomed city. The heated air rising made an enormous suck. Thus did the fire of itself build its own colossal chimney through the atmosphere. Day and night this dead calm continued, and yet, near the flames, the wind was often half a gale, so mighty was the suck.”

In the next section we estimate the smoke from these World War II city fires, the effects on solar irradiance and radiative forcing, and the potential climate impact. Then we compare those estimates to global surface air temperature observations and to climate model simulations.

2 Uncertainties

All these records have large uncertainties. We have good data on the cities that burned, but not on their fuel loading, smoke emissions, or plume heights. Available solar irradiance observations are from two sites only and, although taken on mountaintops, still are affected by the troposphere, thin clouds, and potential errors in the instruments. Sea surface temperature observations during the period of World War II were affected by changing observing techniques and locations, necessitating the use of only surface air temperatures over land. The 1940–1942 El Niño did not produce global warming because of a circulation response that produced cooling over land (Brönnimann et al., 2004), while the La Niñas of 1950 and 1956 did produce global land cooling.

3 Smoke Emissions

Toon, Turco, et al. (2007) provided a method for estimating the smoke emissions from burning cities, taking account of the amount of fuel per capita and integrating over population densities in different cities, and assuming the area burned in each case would be the area of Hiroshima that burned (13 km2) as a result of the atomic bombing on 6 August 1945. Here, because for Japan we know the total area burned, we simplify the approach as
urn:x-wiley:2169897X:media:jgrd54935:jgrd54935-math-0001(1)
where M is the total mass of soot injected into the lower stratosphere (kg), A is the total area burned (m2), F is the total fuel per unit area (kg/m2), E is the percent of fuel that is emitted as soot into the upper troposphere (%), R is the fraction that is not rained out (%), and L is the fraction lofted from the upper troposphere into the lower stratosphere, either injected directly or by subsequent solar heating (%). Caidin (1960) described how the B-29 Superfortress bomber was created in secret as a high-altitude (over 30,000 ft. [9,100 m]), long-range bomber (over 5,000 km) for use by the United States in World War II. Caidin (1960) reported that in Japan, starting with the firebombing of Kobe on the night of 3 February 1945 through the atomic bombing of Nagasaki on 9 August 1945, “the Superfortresses had gutted a total of 178 square miles [461 km2] in 69 cities,” which included the destruction of over 50% of Tokyo, Yokohama, and Kobe (Figure 1). Figure 2 shows part of Shizuoka, as an example, where 66% was destroyed by incendiary bombs. Toon, Turco, et al. (2007) reported a range of values for F and E depending on the city and year but assumed a value of 1.6 kg/m2 for F for Hiroshima and 1.6% for E. They also estimate R to be 80%.
Details are in the caption following the image
Extent of destruction by U.S. bombing of major Japanese cities in 1945 (from United States Military Academy History Department Atlas, http://www.usma.edu/history/SiteAssets/SitePages/World%20War%20II%20Pacific/ww2%20asia%20map%2051.jpg).
Details are in the caption following the image
Part of Shizuoka after it was firebombed on 19 June 1945, (http://www.japanairraids.org/wp-content/uploads/2010/10/A3823.jpg).

As Toon, Turco, et al. (2007) explained, for fires with a diameter exceeding the atmospheric scale height (about 8 km), pyro-convection would directly inject soot into the lower stratosphere. And while we know of no observations of this from the World War II fires, Fromm et al. (2010, 2008) have described the 2001 Chisholm forest fire in Alberta, Canada, which was observed to directly inject about the same amount of smoke into the stratosphere as expected from one Hiroshima-sized firestorm. Fromm et al. (2010) described numerous other such natural fires. Although they were scattered in time and space and did not have as much fuel, they still produced small stratospheric injections. However, even if the initial smoke injection from city fires only had enough lofting to get the smoke into the upper troposphere, subsequent solar heating would loft it into the lower stratosphere in the summer in midlatitudes, as calculated by Toon, Turco, et al. (2007) and as observed after the August 2017 pyrocumulonimus injection from British Columbia forest fires (e.g., Ansmann et al., 2018; Peterson et al., 2018). Because the city fires were at nighttime and did not always persist until daylight, and because some of the city fires were in the spring, with less intense sunlight, we estimate that L is about 0.5, so based on the values above, M for Japan for the summer of 1945 was about 0.5 Tg of soot. However, this estimate is extremely uncertain.

For Europe, although Hamburg (July 1943), Darmstadt (September 1944), Dresden (February 1945; Vonnegut, 1969), and Dortmund (March 1945) were destroyed by firestorms, much of the destruction in Europe was from high-explosive bombs and not fires. The total area in Germany bombed in the entire war was an estimated 205 km2 (Murray & Millett, 2000). The smoke emissions were also spread out over 6 years, starting with the Blitz raids on England in 1940 through the final bombings of Germany ending in Lübeck (April 1945). Fifty-eight percent of the total bombing tonnage on Germany took place in 1944, 23% in 1945, and only 19% in the preceding years http://humanities.exeter.ac.uk/media/universityofexeter/collegeofhumanities/history/researchcentres/centreforthestudyofwarstateandsociety/bombing/THE_BOMBING_OF_GERMANY.pdf. With the same assumptions as for Japan, the smoke emissions from Germany from equation 1 would amount to about 0.1 Tg of soot emitted to the stratosphere in 1944 and an additional 0.05 Tg in 1945.

4 Impacts on Solar Irradiance

As discussed by Hoyt (1979), the Smithsonian Astrophysical Observatory (then abbreviated APO) Solar Constant Program (Aldrich & Hoover, 1954) maintained two long-term observatories of the solar constant (now called solar irradiance) at Mt. Montezuma, Chile (22°40′S, 68°56′W; Figure 3) and Table Mountain, California, USA (34°22′N, 117°41′W; Figure 4). While they adjusted their measurements for Rayleigh scattering, clouds, and water vapor, and tried to remove polluted days,

Details are in the caption following the image
The Mt. Montezuma, Chile (Smithsonian Institution Archive. Image # 2003-19480) solar irradiance observatory.
Details are in the caption following the image
Table Mountain, California (Smithsonian Institution Archive. Image # MAH-21248B) solar irradiance observatory.

“it appears that the influence of aerosol scattering was not entirely removed from the derivation of the solar constant values. Throughout the APO program the solar constant reduction scheme never handled the problem of aerosols or volcanic dust properly.” (Hoyt, 1979, p. 440).

Hence, we might expect this record to include changes of stratospheric aerosols. Indeed Aldrich and Hoover (1954) and Hoyt (1979) commented on the reductions in solar irradiance in 1932 because of volcanic eruptions in Chile.

Figure 5 shows the solar irradiance observations for 1923–1953. The effects of the 10–11 April 1932 Quizapu volcanic eruption in Chile are clear, as are the effects of “solar dimming” from tropospheric pollution beginning around 1950 (Wild, 2009).

Details are in the caption following the image
Monthly average solar irradiance and sunspot number observations. Data from Table 17 in Aldrich and Hoover (1954). The effects of the 10–11 April 1932 Quizapu volcanic eruption in Chile are clear, as are the effects of “solar dimming” from tropospheric pollution beginning around 1950. The purple line is the background solar irradiance during 1940–1943 (1358.4 W/m2).

The lowest values during World War II were in 1944 and 1945, with solar irradiance about 3 W/m2 below the background values, defined as the average for 1940–1943 (1,358.4 W/m2), indicated as a purple line in Figure 5. Accounting for a planetary albedo of about 0.3 and the spherical shape of Earth, this would amount to a radiative forcing of about −0.5 W/m2.

Figure 5 also shows observations of sunspot numbers for the same period. Aldrich and Hoover (1954) suggested that part of the decadal variations of solar irradiance were due to changes in solar emission correlated with the sunspot number. However, Kopp and Lean (2011), using modern satellite observations of total solar irradiance, found that solar irradiance varies by about 1.6 W/m2 between solar maximum and solar minimum, independent of the sunspot value at solar maximum. Since the observations of solar irradiance in Figure 5 vary by 10 W/m2 between solar maximum and solar minimum, the effect of sunspots cannot explain more than a small amount of that variation. In addition, the solar irradiance and sunspot numbers are not correlated for the first part of the record.

An alternative explanation for the reduction of insolation in 1944 and 1945 is tropospheric pollution from industrial activity. However, U.S. industrial production peaked in February 1944, stayed relatively flat until February 1945, and then plummeted during 1945 to a low in October 1945 (Board of Governors of the Federal Reserve System (U.S.), 2017). The period February–October, 1945, was labeled as a recession. By the end of 1945, U.S. industrial production had fallen to 2/3 of its value in 1944. Similarly, industrial production in Germany (Grayling, 2006) and Japan (Caidin, 1960) was drastically diminished by bombing by 1945. Tropospheric aerosols have an e-folding lifetime of only about 1 week. So their effect would have been much larger in 1944 than 1945, yet we found a larger insolation reduction in 1945 (average solar insolation of 1356.4 W/m2) as compared to 1944 (average solar insolation of 1357.0 W/m2). Most likely, the reduction in 1944 was partially from tropospheric aerosols, but in 1945 may have been from stratospheric soot.

5 Temperature Observations

Sea surface temperature observations during the period of World War II and afterwards had several problems. Observers, to avoid being attacked by using lights on deck at night after bringing up a bucket of water to measure, would bring the buckets inside the cabin, which would artificially warm the water. In addition, there was a switch from bucket to intake temperatures soon after the war and there were different shipping patterns during the war, all of which contributed to a large drop from 1945 to 1946 of about 0.3 K (Figure 6), which does not represent the actual temperature variation (Thompson et al., 2008).

Details are in the caption following the image
(a) Global and (b) Northern Hemisphere (NH) annual average temperature anomaly (K) with respect to 1940–1944 mean. Shown are combined ocean and land surface air temperature from National Climatic Data Center (NCDC) Merged Land-Ocean Surface Temperature Analysis (Smith et al., 2008), HadCRUT4 Reanalysis (Morice et al., 2012), and GISTEMP (GISTEMP team, 2016; Hansen et al., 2010), and ocean-only temperatures from the Hadley Centre Nighttime Marine Air Temperature (NMAT; Kent et al., 2013). GISTEMP = Goddard Institute for Space Studies Surface Temperature Analysis.

Because of these issues, we decided to use only land surface air temperatures in our analysis. Figure 7 shows annual-average global mean or Northern Hemisphere (NH) surface air temperature only over land from the two standard analyses from the Goddard Institute for Space Studies (GISTEMP team, 2016; Hansen et al., 2010) and the Climatic Research Unit (Jones et al., 2012; Osborn & Jones, 2014), and both show a drop in 1945 of about 0.1 K in global average temperature, and 0.2 K in NH average temperature, with about half that drop in 1946.

Details are in the caption following the image
(a) Global and (b) Northern Hemisphere (NH) annual average land surface air temperature anomaly (K) with respect to 1940–1944 mean. Data are from CRUTEM (Jones et al., 2012; Osborn & Jones, 2014) and GISTEMP (GISTEMP team, 2016; Hansen et al., 2010). The green whisker (plotted at 1948 in (a)) is the uncertainty of the GISTEMP observations (95% confidence limit) accounting only for incomplete spatial sampling (Hansen et al., 2010). CRUTEM = Climatic Research Unit TEMperature data set. GISTEMP = Goddard Institute for Space Studies Surface Temperature Analysis.

However, when we break up the record into seasonal rather than annual averages, it is seen in Figure 8 that the cooling in 1945 occurred at the beginning of the year, before any smoke would have been injected into the stratosphere. While this cooling in early 1945 may have been due to natural variability, the clear annual signal for 1945 cannot have been caused solely by stratospheric smoke. There were no large volcanic eruptions and no El Niño or La Niña in 1945 or 1946. The other large temperature drops in the record are centered on 1950 and 1956, both periods associated with two of the largest La Niñas on record. The smaller El Niños and La Niñas earlier in the record do not show any impact on the observations, probably because of a combination of their smaller size and poorer observations during that period. As Brönnimann et al. (2004) and Brönnimann (2005, 2007) pointed out, the 1940–1942 El Niño caused cooling over Europe, which cancels out other land warming, so there is no signal of this El Niño in Figures 6-9.

Details are in the caption following the image
(a) Global and (b) Northern Hemisphere seasonal average land surface air temperature anomaly (K) with respect to 1940–1944 mean. Data are from CRUTEM (Jones et al., 2012; Osborn & Jones, 2014) and GISTEMP (GISTEMP team, 2016; Hansen et al., 2010). Vertical red and blue lines indicate El Niño and La Niña events, reconstructed using the Multivariate ENSO Index (units on right axes, Wolter & Timlin, 2011). ENSO = El Niño Southern Oscillation; CRUTEM = Climatic Research Unit TEMperature data set; GISTEMP = Goddard Institute for Space Studies Surface Temperature Analysis. Abscissa tick marks are for January of each year.
Details are in the caption following the image
(a) Global and (b) Northern Hemisphere seasonal average land surface air temperature anomalies, with respect to the average for 1940–1944, from the CMIP5 historical simulations (Table 1), and observations from CRUTEM (Jones et al., 2012; Osborn & Jones, 2014) and GISTEMP (GISTEMP team, 2016; Hansen et al., 2010). The thin gray lines are the first three ensemble members for each separate climate model (Table 1), and the thick black line is the multimodel mean. Vertical red and blue lines indicate El Niño and La Niña events, reconstructed using the Multivariate ENSO Index (Wolter & Timlin, 2011). CMIP5 = Coupled Model Intercomparison Project 5; GISTEMP = Goddard Institute for Space Studies Surface Temperature Analysis; ENSO = El Niño Southern Oscillation. Abscissa tick marks are for January of each year.

Natural climate variability might be expected to prevent the detection of small forced climate responses. To evaluate this, we compared the observations to 24 Coupled Model Intercomparison Project 5 (Taylor et al., 2012) simulations (see Table 1) of historical climate change, forced by observed climate forcings, but not soot from World War II city fires. The results are plotted in Figure 9 and show no mean temperature drops during the period. Because the model simulations had varying numbers of ensemble members, to give them all equal weight, we only used the first three ensemble members for each model. Since simulated El Niño and La Niña in the models would not be at the same time as those observed, these would randomly cancel out in the average. But we removed the additional variance produced by those El Niño Southern Oscillation (ENSO) events using a pointwise linear regression and show the resulting curves in Figure 9. This made little difference in the results. The standard deviations of seasonal temperatures from the model runs were 0.28 K for global means without ENSO removal, 0.28 K for NH means without ENSO removal, and 0.25 K (global) and 0.27 K (NH) with ENSO removal. Thus, the observed small cooling of about 0.2 K in 1945 could have easily been by chance. Standard deviations for annual anomalies before ENSO removal were 0.12 K and 0.13 K for global and NH, respectively, and 0.11 K for global and 0.12 K for NH after ENSO removal.

Table 1. Coupled Model Intercomparison Project 5 Models Used
Model Resolution (°) (lat × lon) Vertical levels Ensemble members Reference
ACCESS1-3 1.25 × 1.875 38 3 Bi et al. (2013)
BCC-CSM1-1 2.8 × 2.8 26 3 Wu et al. (2013)
BCC-CSM1-1-M 1.11 × 1.125 26 3 Wu et al. (2013)
CanESM2 2.8 × 2.8 35 5 https://www.canada.ca/en/environment-climate-change/services/climate-change/science-research-data/modeling-projections-analysis/centre-modelling-analysis/models/second-generation-earth-system-model.html
CESM1-CAM5 0.94 × 1.25 26 3 Hurrell et al. (2013)
CESM1-FASTCHEM 0.94 × 1.25 26 3 Hurrell et al. (2013)
CCSM4 0.94 × 1.25 26 8 Gent et al. (2011)
CNRM-CM5 1.4 × 1.4 31 10 Voldoire et al. (2012)
CSIRO-Mk3-6-0 1.865 × 1.875 18 10 Rotstayn et al. (2010)
GFDL-CM2p1 2 × 2.5 24 10 Delworth et al. (2006)
GFDL-CM3 2 × 2.5 48 5 Donner et al. (2011)
GISS-E2-H 2 × 2.5 40 18 Schmidt et al. (2014)
GISS-E2-R 2 × 2.5 40 18 Schmidt et al. (2014)
HadCM3 2.5 × 3.75 19 10 Johns et al. (2003)
HadGEM2-ES 1.25 × 1.875 38 4 Collins et al. (2011)
IPSL-CM5A-LR 1.9 × 3.75 39 6 Dufresne et al. (2013)
IPSL-CM5A-MR 1.27 × 2.5 39 3 Dufresne et al. (2013)
MIROC-ESM 2.8 × 2.8125 80 3 Watanabe et al. (2011)
MIROC5 1.4 × 1.4 40 5 Watanabe et al. (2010)
MPI-ESM-LR 1.865 × 1.875 47 3 Giorgetta et al. (2013)
MPI-ESM-MR 1.865 × 1.875 95 3 Giorgetta et al. (2013)
MPI-ESM-P 1.865 × 1.875 47 2 Giorgetta et al. (2013)
MRI-CGCM3 1.12 × 1.125 48 5 Yukimoto et al. (2012)
NorESM1-M 1.9 × 2.5 26 3 Bentsen et al. (2013)
  • Note. The total number of ensemble members available from each model is listed, but we only used the first three from each, except for MPI-ESM-P, which only had two.

6 Discussion and Conclusions

Robock, Oman, Stenchikov, et al. (2007) found a reduction of downward shortwave radiation at the surface of about 13 W/m2 1 year after an injection of 5 Tg of soot, while Mills et al. (2014) and Stenke et al. (2013) found a reduction of about 9 W/m2 due to faster immediate soot removal than Robock, Oman, Stenchikov, et al. (2007). One tenth of this soot (our estimate of the stratospheric loading in 1945) would produce a reduction of about 1 W/m2, quite a bit more than calculated from the data in Figure 5. Assuming the observations were correct, our estimates of smoke injection from the World War II fires were probably too large.

Robock, Oman, Stenchikov, et al. (2007) calculated the impacts of a 5-Tg soot injection into the upper troposphere and found global average surface air temperature reduction of about 1 K after 1 year, with larger cooling over land. They also simulated an injection of 1 Tg and found with these low levels of loading, the climate response was essentially linear. Therefore, the injection of 0.5–1 Tg of soot into the upper troposphere from city fires during World War II would be expected to produce 0.1–0.2 K global average cooling, with larger cooling over land and over the NH (because the NH has a higher fraction of land than the globe). But if less smoke was injected, the expected cooling would have been less. While these results are consistent with the results shown in Figures 6 and 7, when examining the observed signal further and comparing them to natural variability, it is not possible to detect a statistically significant signal.

We reject the hypothesis that tropospheric aerosols were an important cause of 1944 and 1945 temperature changes. There were more tropospheric aerosols in 1944 than in 1945, and their effect would have been regional and short-lived.

While the results in this paper might be considered “negative,” in that we were not able to make a case that we could observe the impacts of smoke from fires ignited by incendiary weapons during World War II, it is important to document this result. Detection of the signal was not possible because of poor data on smoke emissions, solar radiation, and surface temperature, natural variability, and the small expected signal. Nevertheless, these results do not provide observational support to counter nuclear winter theory as simulated by Robock, Oman, & Stenchikov (2007); Robock, Oman, Stenchikov, et al. (2007); Stenke et al. (2013), and Mills et al. (2014).

Acknowledgments

This work is supported by US National Science Foundation grant AGS-1430051 and a grant from the Open Philanthropy Project. We thank David W. J. Thompson, Phil Jones, and Brian Toon for valuable comments, and Michael Fromm and one anonymous reviewer for additional constructive comments on a previous version of this paper. We thank the Centre for Environmental Data Analysis (http://ceda.ac.uk) for making the CMIP5 output available. The data used in the paper are listed in the references and tables.