Global Atmospheric Effects of Nuclear War Our research on the global effects of nuclear war has led to improved calculations of fire development smoke injection into the atmosphere and the complex interactive transport of smoke in a global climate model However significant uncertainties remain in the chain of calculations that would predict the climatic effects of smoke generated in a nuclear war-the nuclear winter effect For further information contact Michael C MacCracken 415 -122-1826 10 This article reviews the current research program and recent findings of Laboratory scientists investigating possible global-scale climatic effects from material principally smoke that would be injected into the atmosphere as a consequence of a nuclear war Our research addresses the number and nature of possible fires from a nuclear war the amount of smoke that could be produced how it would be injected into the atmosphere and spread over broad areas the processes that determine the optical properties and lifetime of smoke in the atmosphere and the influence of varying amounts of smoke on weather and climate In the first section below we give a brief historical account of the recognition of the more important global effects of nuclear war over the past forty years We emphasize the recent 1982 recognition of possible serious climatic consequences of large amounts of smoke in the atmosphere we also outline the major uncertainties in our understanding of this effect that make further research necessary Subsequent sections deal with our research on fire smoke and climate G lobal Effects of Nuclear War Some History Radioactive Fallout In the event of a nuclear war the most violent and destructive effects of blast heat and prompt nuclear radiation although they would produce unparalleled devastation would nevertheless be localized in the areas immediately surrounding the explosions For many years radioactive fallout was the only significant effect that was recognized to extend well beyond the explosion area There was concern about possible local fallout danger for even the first nuclear explosion the Trinity test at Alamogordo cw Mexico in July 1945 In fact some small but measurable effects of radioactivity from Trinity were observed more than a thousand miles from the test The development of megaton weapons in the early 1950s increased by more than a hundredfold the potential danger of fallout from individual explosions The 15-Mt Bravo test in 1954 at the Bikini Atoll produced D EFE SE PROGRAMS potentially lethal lev ls of fallout mor than 2-W ·m do ·mdnd of the pl ion The ri ing 1reball of a nuclear '- Pl si n a or n ar th earth' urface uc · up vast quantiti of dirt and du t and the radioactive vapors of the pl ion readily conden on this debri which ubsequently falls out 0 enerall ' dowm nd of the e plo ion uch urface urs produce what is tenned local fallout ln a nudear war in 'ohin numerous urface or nearsurface burs large numbe of radiation ca·ualti would be expected from local fallout ·uclear e losions detonated hi h en u h so that the fireball does not touch the urface are called air bursts and produce little local fallout However the radioactivity of an air-burst explosion and about one-half that produced by a urface e -pJ ion will be carried into the upper troposphere and if the yield is large enough into the tratosphere This radioactivity i carried by the wind and in time preads throughout the atmosphere Eventually variou proc lead to the continuing global deposition on land and ocean of the portion of this material that has not radioactively decayed in the atmo phere Extensive research on fallout mechanisms and on the possible biological consequences of fallout radiation has been carried out ince the l 9- -l Bravo test in whjch Marshall I landers and Japanese fishermen were eriously irradiated Recent Laboratory research on global fallout i discu ed in the box on p 14 Ozone Depletion and Dust In the early 1970s a new global effect wa recognized by L L scientists The large amount of oxides of nitrogen produced in a nuclear fireball could be carried into the tratosphere where these oxide would catalyze reactions that would deplete ozone concentrations In a nuclear war v ith many high-yield explosions the oxides of nitrogen would tend to spread latitudinally in the tratosphere Over much of the earth this would lead to a decrease of the layer of ozone that protects the earth from mo t of the sun' ultraviolet radiation A 1975 study conducted by the ational cademy of Sciences entitl d the LongTerm Worldwide Effect of ultiple 1 1uclear-Wea pon Detonation '' 1 valuat d the eff ct of a 10 000- t war on the oz ne layer widespread 50% ozone depl tion lasting for a year or more was e timated with biological con equenc timated to b more vere than global depo ition of radiation Several Laboratory cienti t had key rol in thi review 2 and research on the ubject has continued to the pr ent at L L The 1975 Academy tudy also con idered po ible global climatic effects from large amounts of dust that could be injected into the trato phere in a nuclear war Jt wa concluded by analogy with the near-absence of global climatic eff cts following large volcanic eruptions that du t would not produce ignificant effects Ozone and du t effects are discu ed further in the box on p 14 Smoke ln 19 2 the Royal Swedi h Academy of Sciences ponsored a new inquiry into the consequences of nuclear war Paul Crutzen and John Birks experts on air chemistry were asked to review the pas ible atmospheric effects of a war In the course of ree aluating effects uch a nitrous oxide production ozone depletion in the atmosphere and po ible enhanced ozone production in the troposphere they recognized that smoke and other chemicals injected into the atmosphere from fires started by the nuclear explosions might have major global atmospheric effects previously not con idered quantitatively They suggested that fires in forests and citie as well as burning oil and gas supplies could produce tremendous amounts of smoke They specifically estimated the amount of smoke likely to come from forest fires and noted qualitatively the possible serious effects on atmospheric chemistry and the climate Crutzen and Birks's work was published in the journal Ambia in the spring of 1982 3 At about the same time another group of scientists Richard Turco Brian Toon Tom Ackerman James Pollack and Carl Sagan were undertaking a reassessment of the findings of the 1975 11 Academy study When they heard of Crutzen and Birks's considerations of the possible effects of smoke they expanded their study to include the first quantitative assessments of the global climatic effects of large amounts of smoke They circulated an early version of a report on their work in the spring of 1983 and in December 1983 published a final version in the magazine Science entitled Nuclear Winter Global Consequences of Multiple Nuclear Explosions 4 This paper is referred to as the TIA PS report an acronym formed from the authors' initials The TIA PS report conjectures that a nuclear war involving over 10 000 explosions with a total yield of 5000 Mt could produce and inject into the atmosphere over 200 Tg 200 million tonnes of smoke Spread over the northern hemisphere the smoke would absorb most of the sun's energy high in the atmosphere Temperatures over major land areas could drop by about 35°C and remain below freezing for more than a month If these predictions are correct the phrase nuclear winter is an apt one for the effect A companion report to the TIA PS study in the same issue of Science5 by twenty scientists mostly biologists and ecologists attempted a first assessment of the biological consequences of nuclear winter It concluded that in addition to the direct effects of a nuclear w ar extended subfreezing temperatures and low light levels would have further catastrophic effects and that the population size of Homo sapiens conceivably could be reduced to prehistoric levels or below and extinction of the human species itself cannot be excluded The TTAPS study required a chain of uncertain assumptions concerning the nature of a nuclear war how much smoke would be generated how high it would rise how long it would remain in the atmosphere and how effectively it would absorb sunlight Furthermore the estimates of temperature change were based on results from a highly simplified model of the global climate Their onedimensional radiative-convective model treats only the vertical structure of the 12 atmosphere it neglects the land and ocean difference horizontal spread of smoke diurnal and seasonal varia tions of solar radiation and many other factors By the spring of 1983 many atmospheric scientists were aware of Crutzen and Birks's work and had seen early versions of the TIA PS report Several research groups undertook to improve on the TIAPS analysis by doing calculations using more realistic global climate models These groups included the National Center for Atmospheric Research NCAR in Boulder Colorado the Computing Center of the Academy of Sciences of the Soviet Union in Moscow the Atmospheric and Geophysical Sciences Division at LLNL and the Los Alamos ational Laboratory To date new results have mainly been obtained in the areas of global spread of smoke and the consequent climatic effects For the new investigations the researchers retained most of the original assumptions of Crutzen and Birks and of TIA PS concerning how m uch smoke could be generated what kind and what would be the early-time phenomena Under these assumptions the new results although differing sligh tly in detail from one another agree qualitatively with the basic findings of the TTAPS study Specifically if a sufficient amount of black smoke is injected high into the atmosphere during a warm part of the year then significant climatic cooling of land surfaces can occur The average cooling over land calculated in the more realistic global models is significantly less than the maximum cooling obtained in the TTAPS work although the temperature changes over the centers of continents w ere similar to those of TIA PS For a given temperature change the newer calculations tend to give higher fma1 temperatures than did TIA PS because the assumed initial temperatures in TIA PS were substantially lower than typical midcontinentaJ summer values The qualitative agreement among these results is somewhat deceptive however not only because the new work starts with the same assumptions about smoke generation and injection as did TTAPS but because much of the global DEFENSE PROGRAMS circulation phenomenology itself needs to be reexamined in the light of the new conditions obtained with injection of large amounts of make into the upper atmosphere hat is clear is that the early work of Crutzen and Birks and of TTAPS has raised a serious question Obtaining even approximate answers to it ·will take much more extensive research A broad review of the predictions of Crutzen and Birks of the TIAPS group and of ubsequent research findings by others has been carried out by the ational Academy of Sciences and was announced in December 1984 6 The new Academy report confirms the possibility of a nuclear winter It also highlights the substantial range of uncertainty in the predictions and the need for further research Ti he Laboratory's Global Effects Program The Laboratory has been involved in research related to the global effects of nuclear ar for more than 25 years In early 1983 we initiated exploratory tudies of the nuclear winter effect following publication of the Crutzen and Birks article and the draft TTAPS study A formal research program on this topic was begun at LLNL in late 1983 funded primarily by the Laboratory's defense programs The elements of our program are • To estimate plausible ranges for the amount of material that might be ignited • To estimate the development intensity and spreading rate of fires • To calculate under various conditions the height to which fire plumes might rise and the vertical distribution of injected smoke • To determine the evolution of the physical chemical and radiative characteristics of smoke in the rising plume and as the particles interact with water vapor clouds and rain • To calculate the effect of smoke particles on solar and infrared radiation • To use atmospheric models to calculate the spreading and scavenging of smoke from plume to global scales and the induced effects on weather and climate • To develop an overview of the potential biological and ecological con equences of a major exchange Even the broad range of topics included in the LLNL research program does not adequately address many areas of the problem Thus we are also involved in cooperative efforts with the Defense uclear Agency D A ln addition we have assisted the ational Oimate Program Office in developing a national plan to augment the efforts now under way by the DOE and the DNA 7 With several years of research we believe that the major uncertainties in the predictions can be significantly reduced S cenarios The predicted cooling of land surfaces is highly dependent on the amount of smoke that would be generated in a nuclear war For total quantities of smoke less than about 50 Tg the cooling effects would probably be primarily regional rather than hemispheric or global Fifty teragrams is about ten times the current annual emission of smoke from forest fires in the U S For more than about 300 Tg of smoke it is likely that there could be climatic effects darkness and low temperatures over much of the northern and southern hemispheres Adding much more smoke would not make the effects much more severe but would probably prolong them For quantities of smoke in the range from about 50 to 300 Tg spanning the values considered by TTAPS the predicted temperature changes are strongly dependent on the amount and characteristics of the smoke the height at which it is injected the rate of spread the time of year and many other factors Each of these factors depends critically on what kind of nuclear exchange is assumed the combustible nature of the targets and how much fuel would bum To deal with these problems a series of hypothetical nuclear wars or scenarios are being constructed to span possible ranges Previous studies have made rather arbitrary assumptions in developing baseline nuclear war scenarios For 13 example Crut2-en and Birks used a scenario prepared by advisors to the journal Ambio 3 This group constructed w hat they refer to as a limited scenario including nearly 15 000 warheads totaling about 5750 Mt In this scenario all cities of NATO and Warsaw Pact countri es with populations of over 100 000 people and all other cities with popula tions greater than 500 000 were targeted with a total of about 5000 warheads on about 1100 cities In addition they assumed that about 6600 warheads would be directed to other Radionuclides Nitrogen Oxides and Dust One of the most widely known effects expected from a nuclear war is fallout-the deposition of radioactive material produced by nuclear explosions In addition to fallout the interaction of clouds of radioactive debris with precipitation systems can result in the localiZ-ed deposition of radioactive material a process termed rainout 8 Relatively intense fallout or rainout could occur adjacent to and downwind of targets subject to nearsurface nuclear explosions The most likely targets for surface bursts are hardened missile sites and command centers Protection from intense local fallout and rainout requires personnel to be evacuated from the affected areas or to remain in adequate shelters for several weeks N uclear explosions occurring well above targets are called air bursts Such explosions would create the most widespread fires and blast damage The radionuclides produced would tend to be injected high in the atmosphere and dis persed over large areas Except in regions w here relatively fresh debris is carried to the ground by rain in precipitation systems the fallout radiation doses would likely remain well below critical levels especially if protective actions were taken 9 In the early 1970s scientists at LLNL and elsewhere recognized the potential global effects on stratospheric ozone of the nitrogen oxides that are created in nuclear fireballs Explosions of about 0 5 Mt or larger can loft nitrogen oxides into the stratosphere where they could interact chemically to reduce the concentration of ozone Stratospheric ozone absorbs a large fraction of the ultraviolet radiation emitted by the sun thereby protecting the surface of the earth from excessive exposure A substantial reduction in ozone would allow increased ultraviolet radiation to reach the surface which could in turn lead to severe sunburn in a short time 10 and damage to unprotected plants and animals Adequate clothing or shelter would protect humans The amount of injected nitrogen oxides and the resulting extent of ozone depletion depends on the yield and number of explosions and thus on the character of the war that is assumed For a war scenario involving about 10 000 Mt of high-yield weapons estimates made in the 1970s indicated that stratospheric ozone would be 14 reduced by up to 50% and that the effect would last for several years However the trend to warheads with smaller yields has led to lower estimates of oxides injected into the stratosphere and consequently lower ozone depletion In addition because smoke and dust from the explosions may block much of any increased ultraviolet radiation estimates of the potential exposure at the surface have been further reduced The 1982 study of Crutzen and Birks3 suggested that the presence of large amounts of nitrous oxide in the troposphere could actually favor chemical processes leading to the production of ozone Their studies indicated that in the absence of smoke smog levels typical of Los Angeles might develop Subsequent investigations at LLNL suggest that smoke would indeed prevent the photochemical reactions that produce smog and that the essential ingredients nitrous oxides and hydrocarbons that might create smog m ay be scavenged more rapidl y than the smoke particles 11 By the time that solar radiation would again be intense enough to induce smog it is likely that the smogproducing chemical ingredients would have largely disappeared The climatic effects of dust that would be injected into the atmosphere as a result of many surface bursts have been a concern for many years The National Academy of Sciences study of 1975 1 briefly examined the potential climatic effects of dust injection and concluded that the effects would not be significantly different from those following major volcanic eruptions such as Tambora in 1815 or Krakatoa in 1883 Although eruptions during the past may well have caused unusual and cooler weather they have not except near the eruption seriously affected human plant or other animal life It has been suggested that Tambora may have induced what has been called the year without a summer in the northeastern U S and Europe 12 The year was characterized by locally serious crop failures due to unseasonal frosts and snowfalls however the effects were small in most areas and near-normal weather patterns returned the following year Krakatoa produced temperature decreases of at most a fraction of a degree in hemispheric average temperature DEFE SE PROGRAMS military targe and 3100 warheads 1 ·ould be used on indu trial and energyrelated targe The TI PS ba line a urned that about 10 -100 expl ions and 5000 1t would be used with the yield distributed among veral types of target They also con ·dered numerous other scenario that ran ed from 100 to 25 000 t The latter number i actual _ about n ice the total ield of the approximately SO 000 weapons now in existence according to the recent ationaJ cademy of Science 'AS report 6 For the Academy study uself it was assumed that half of all existing weapons would be exploded for a otaJ of about 25 000 explosions and 6500 It In uch simplified scenarios many matters of importance tend to be overlooked For instance no rationale or e idence exists for targeting all large cities around the world or for other a pects of the Ambia TfAPS or AS scenarios Multiple weapons on important military targets would reduce the number of fires that might be started The proximity of military targets to developed areas and forests would aJ o alter the number and kinds of fires tarted To gain a better perspective we are working with the D A to develop a range of more detailed scenarios They will incorporate data on the makeup of the U S and Soviet arsenals categories of possible targets and target areas and various practical considerations Our intent is to establish a set of buildingblock scenarios that can be combined to provide a range of inputs to the calculations In addition to developing a range of possible cenarios and target categories we must estimate the amount and type of combustible fuel in each of the target categories The TTAPS study for example assumed continental-average forest coverage around all military targets In fact most U S missile fields are located in fields or prairie which would generate much less smoke than forests Since urban fires are potentially the most important source of smoke we are conducting studies to estimate the fuel loading of typical cities W hat Is Injected into the Atmosphere To predict the climatic effect of a nuclear war we mu t know the amount of moke that is produced by fires the chemical and optical characteristics of the smoke and the altitude at which the make is initially injected into the atmosphere The amount and propertie of smoke depend on the fuel and the conditions under which they bum We must estimate not only the total amount of fuel actually consumed by the fires but also how the fires bum Smoldering fires of low intensity produce relatively large amounts of smoke per unit weight of fuel consumed The smoke remains relatively low in the atmosphere a few kilometres is whitish or gray and acts primarily to scatter light rather than absorb it Flaming fires produce less total smoke but the smoke is blacker and tends to absorb much of the light striking it Intense flaming fires can release large amounts of energy and can loft black smoke into the upper troposphere about 5 to 10 km above the surface of the earth Smoke lofted to these higher altitudes could induce more cooling at the surface of the earth than smoke at altitudes of a few kilometres because the particles would absorb incident solar radiation in the troposphere above the natural greenhouse gases 13 In the unperturbed atmosphere smoke higher in the troposphere would be expected to remain longer than smoke injected nearer the surface because there is less precipitation at higher altitudes Very intense fires may even inject some smoke into the stratosphere where it would remain for even longer times The dividing line between the troposphere and stratosphere is called the tropopause and varies with latitude In our models for smoke injection we have selected a globally averaged altitude of 11 5 km for the tropopause Research on the Initiation and Spread of Fires Of the two types of fire that have been considered in global-effects research more is known about the 15 initiation and spread of fires in • vild lands than in urban areas Crutzen and Birks proposed wild-land fires as a large source of smoke following a nuclear exchange They did not treat urban fires at all although they included an estimate for the burning of fuels stored above ground They suggested that in a nuclear war on the order of a milbon square kilometres of forest would bum in the northern hemisphere emitting about 7% of the weight of fuel as smoke for a total of about 200 Tg of smoke The TTAPS group estimated that about 80 Tg of smoke might be produced from burning forests They assumed the area of forest fires to be half a million square kilometres and a more reasonable smoke emission rate of about 3% ln a recent more detailed analysis of possible targets in forested and other wild-land areas Small and Bush argue that only 200 000 km 2 of wild land might be expected to bum 1 Because much of the region surrounding targets is grassland rather than forest and because the resulting fuel load and emission rates are lower they estimated that at most 2 2 Tg of smoke would result from these sources Others have suggested that this estimate may be somewhat low 15 From measurements of the optical properties of smoke produced from forest fuels we now know that such smoke is less absorbing of solar radiation than was assumed by TTAPS Furthermore the smoke from forest fires may not rise as high into the atmosphere as the TTAPS report assumed Because of these factors we now consider the smoke from wild-land fires to be a much less important contributor to the nuclearwinter effect than smoke from urban fires The amount of smoke that would be produced from urban fires is difficult to estimate Large forest fires do not serve as a useful analogy because they bum quite differently In addition forest-fire fuel loadings are only about one-tenth those of an average suburban area and perhaps one-thirtieth those of a typical city center Detailed current surveys of the amounts of burnable material present in cities are not available The amounts and types of combustible material are 16 expected to be different in ariou countries of interest and even from region to region in each country In addition we do not understa nd how an entire city might bum after a nuclear attack Good estimates of th e amount of smoke produced by a large fire one in which 50 to 250 km 2 of a city are ignited at once are unavailable The TTAPS baseline scenario assumes a total urban area of 230 000 km 2 equivalent to abou 2000 cities the size of San Francisco would be completely burned Many significant questions about smoke from city fires remain unanswered including for example • Whether or not the fuel in the severely blast-damaged area of a city might be so covered with debris that it does not bum m uch if at all • How well the fires are supplied with oxygen • The peak temperatures that are reached • Whether fires are hot enough to ignite neighboring structures by thermal radiation • How much smoke is produced by various fuels particularly the ne er synthetics burned under extreme conditions Laboratory measurements of smoke emissions under differing temperature conditions are available for some fuels However because we cannot yet predict the conditions of a nuclear-explosiongenerated fire they are of limited value To approach the problem of urban fires we have obtained detailed codes that were developed as part of civil defense research programs in the 1960s The codes were used to characterize the burning of cities after a nuclear blast and can be helpful in estimating ignition fire-spread in various parts of a city and how much fuel would be expected to bum These models still have many limitations For example the codes assume that each building of a given type will bum in a prescribed manner In general a fire is assumed to build slowly over about half an hour become a stage-three fire in which the entire building is engulfed in flames in an hour and then die down after roughly another D EFE SE PROGRAMS hour · thou h on were dealing with ar i la ed building Po ible effe that ma be important uch a immediate Jlashowr to naming conditions du to hi h position of energy within an room from the fireball are not treated The civil-defense codes al o do not adequately tr at the problem of whether or not fires tart and pread in tho e areas of a city that are e entially reduced to rubble by bla t It is nece ar_ t0 de ·elop prescription for fire initiation and ·pread in uch area Although the older codes 'ere helpful for civil-defen e a-- men they may require _ubs antiaJ modifications to provide an adequate picture of the kind of urban fire hith which we are concerned D pite the e limitation we have used the codes for ome exploratory calculation If urban fires develop as no modeled and fuel loadings are es than 10 g cm then the rates of energy release will not be high enough to prO de what are believed to be necessary condition for a firestorm such a occurred at Hamburg in 1943 The estimated energy-relea e rate at Hamburg was 1 -l x 105 m 2 over an area of about 12 km 2 We al o have found that the total amount of fuel that ·ouJd be burned the intensity of the fire and the area over which the fire would pread depend on the particular fuel distribution that is as urned We have explored the dependency on fuel distribution and other factors with calculations for a uniform city and for a repre entation of San Jose California The uniform city consisted of identical two-story frame houses equally spaced in identical housing tracts For the basic calculations the average fuel loading was 3 g cm 2 The curve labeled uniform city in Fig 1 shows the cumulative fuel con umption versus time as predicted by the fire-spread code For these calculations a 1-Mt airburst was a sumed The calculation neglects any burning in the heavily blast-damaged region out to about 9 km the 25-kPa level and assumes that fires are ignited by the thermal pulse out to about 13 km where the thermal energy is 325 kJ m 2 The code predicts that about 4 Tg of fuel i burned in the fir t even hours and another 2 Tg in the re t of a 24-hour period In addition to thi ba ic calculation numerou additional calculation were performed varying parameters uch a wind speed building den it fuel loading range of fire brand and frequency of econdar fire cond et of imulations used a representation of the greater San Jose region as it wa in 196 data available from old ci il-defen e tudies to analyze how a more realistic city would bum Again a 1-Mt airburst wa assumed However the area surrounding the burst point wa of relatively low building and fuel density and there were many vacant tract area invol ed A a consequence the potential fuel for burning was much less than for the uniform city calculation Figure 1 include a curve for the cumulative fuel burned with time for the San Jose 1968 calculations The examples of the uniform city and San Jose dramatize how different by more than a factor of ten fuel consumption and implicitly smoke production can be for different city representations and aim-point choices Much more effort is needed in characterizing fuel distribution of urban areas In the coming year we will develop new fuel-loading descriptions for several typical cities These descriptions will enable us to estimate how the total amount of fuel consumed varies with city type so that we can make comparisons with other estimates We 0 5 N 0 Fig 1 - 4 Cl E q 3 0 0 ai 2 2 iii San Jose 1968 5 10 15 Time hours 20 25 The cumulative fuel consumed outside the blast-damaged area of a uniform city is roughly ten times greater than that calculated for the area of northwestern San Jose California as characterized in 1968 Such differences illustrate the need to develop additional and more accurate fuel-loading descriptions which will enable us to determine how the total amount of consumed fuel varies with city type and configuration 17 will also be developing a new model that can describe how blast-damaged areas bum Our intention is to test our model against available observations of pa t fire spread and possible future experiments Although understanding how fires generated by nuclear explosions really bum and how much smoke they actually produce are difficult problems an approach that combines both experiments and models should help to address these two important issues The national research plan recommends such an approach Plume-Rise and SmokeInjection Altitude The height to which smoke is lofted by the heat of a fire is important because smoke lofted to high altitudes is expected to have a much longer lifetime than smoke injected at altitudes closer to the earth's surface U all the smoke from fires were injected at low altitudes below about 2 km not only could it be removed relati ely quickly by rainfall but also only very limited cooling or even an increase in surface temperature would be expected while the smoke remained The amount of light reaching the surface could nonetheless still be substantially reduced Several groups have investigated the atmospheric dynamics associated with the sm oke plumes from intense fires those with energy-release ra tes of about 1 5 W m 2 and greater Two recent papers relied on an analytical model tha t accurately describes the rise of smallscale plumes to estimate the height of smoke injection for the much larger fires in w hich we are interested 16 17 Although the estimates appear to be reasonable it is not known how accurate they are For the TTAPS study the heights to which smoke might be carried above a fire were not explicitly calculated The authors assumed a range of injection altitudes between 1 and 5 km for forest fires and between 1 and 7 km for urban fires Five percent of the urban fires were assumed to be firestorms that lofted smoke from 5 to 19 km Penner et al 18 have applied a numerical h ydrodynamic model to calculate the heights to which smoke 18 could be lofted Since our effort began other groups using similar models have obtained results that are consistent with our imulations for very large fires 19 r0 models to date adequately address the significant issue of the extent to which rain forming in the convection column cavenges the make as it rises and preads Using our model to simulate a h pothetical line-source that is twodimensional fire we can investigate the influence of a variety of factors such as fire intensity condensation of water vapor and background wind speeds see Fig 2a-c We are currently investigating some of these factors with a version of the code that allows the more realistic simulation of a fu three-dimensional fire The three-dimensional code is relatively expensive to run and the results shown in Fig 2d tend to confirm those obtained with the two-dimensional code In particular • For intense fires with energy-release rates of 8 9 x 104 W m 2 over a 5-kmradius circle that may be typical of urban centers most of the smoke is lofted to altitudes between 5 and 10 km See Fig 3 for comparison of our results with the TTAPS distributions for comparable amounts of smoke For these fires significant amounts of water vapor may condense raising the possibility of early scavenging of smoke depending on its chemical and physkaJ characteristics see below • For fires of medium intensity with energy-release rates of 1 4 x 10 W m 2 assumed typical of suburban areas most of the smoke is lofted to altitudes between 2 and 8 km • For fires of low intensity with energy-release rates of 2 3 x 103 W m 2 typical of forest fires most of the smoke remains below 3 km • The smoke directly above a fire reaches somewhat higher altitudes than the altitude to w hich most of the sm oke eventually settles an d disperses • Background winds tend to limit the upward movement of smoke thereby decreasing the peak altitude of injection • Water condensation within a plume releases latent heat and can add buoyancy that will cause the smoke to DEFENSE PROGRAMS 20 - - - - -- -- -- - -- -- - -- - - - a Two-dimensional model with no water and no wind 10 ----- 5 10 2 94 - - -- 6 10 ' 10 ----___ 20 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - b Two-dimensional model with water but no wind 10 2 88 X 10 5 76 X 10 10 5 g g 1 73 X 10 - 5 E 10 i 20 30 so 40 ct 20 r - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - c Two-dimensional model with both water and wind 3 52 7 05 1 76 10 X X 10 - 6 10 - S 10 S Fig 2 Simulations of a high-intensity fire 8 9 x 104 W m 2 allow us to determine how smoke is injected into and spreads through the atmosphere The contours shown here as solid lines indicate the maximum mass-mixing ratios of smoke expressed in units of grams of smoke per gram of background air The interval between each contour is one-tenth of the maximum calculated for a given simulation a Smoke contours are shown for the case without condensing water vapor or background wind b When a global-average relativehumidity profile is assumed water vapor carried aloft in the convection column condenses at high altitude and latent heat is released to drive the smoke higher In both a and b only one-half of the smoke profile is shown because these two-dimensional simulations are symmetric about the origin of the smoke c Background wind added to the two-dimensional simulation with water carries the smoke to the downwind side of the source region d For a three-dimensional simulation both condensing water vapor and background wind are assumed to be present With background wind the smoke lofted in our two-dimensional model does not reach an altitude that is as high as that for the more realistic three-dimensional case Note that the horizontal scales in both c and d are shifted to the right to show asymmetric smoke profiles about the origin 20 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - d Three-dimensional model with both water and wind 10 Maximum 3 98 0 10 - 5 g g L- l -_L 1 SU --L l l l l ll lJ_L _L_ L___t_ L_L _L____J __J_ l_L___L_ J_ J ____L____J __J__J - 10 0 10 20 30 40 Downwind distance km 19 Fig 3 The integrated smoke density is shown as a function of altitude for each of our four simulations for high-intensity fires depicted in Fig 2 solid lines Depending upon background conditions most of the smoke from an intense fire is lofted to altitudes between 5 and 10 km In general the lower-altitude injections of smoke that were assumed by TTAPS 4 see shaded reg ion below ab out 7 km correspond quite well to our patterns for low- and mediumintensity fires which are not shown in this figure Our simulations of a highintensity fire indicate that little smoke is lofted above the tropopause dashed hori zontal line and into the stratosphere Note the three-dimensional model curve is plotted ten times less tha n its actual values to permit compari son with the other curves be lofted from 2 to 3 km higher than the peak altitudes calculated without considering water condensation • Injection heights thus can vary with the amount of water vapor present in the atmosphere in the area of fires Our calculated altitudes for smoke injection from fires of various sizes are generally consistent with those assumed for urban fires in the TTAPS study Our results do indjcate that lofting of smoke into the stratosphere is unlikely unless the background atmosphere is rughly unstable and the fires are extraordinarily intense as was the case in Hamburg Radiative Properties of Smoke The railiative properties of smoke particles depend on three important physical characteristics the composition the size and the shape of the particles The composition of particles affects their scattering and absorption properties Particles with a high fraction of elemental carbon appear black and are effective absorbers of sunlight Particles with a low fraction of elemental carbon are lighter in color and tend to scatter sunlight more than they absorb it 20 - - - - - - - - -- - -- -- -- - - - - - - - -- - - Two-dimensional model with water but no wind Two-dimensional model with both water and wind Three-dimensional model with both water and wind 15 Two-dimensional model -- ___ --•Hh oo wore _• - 5 TTAPS4 0 L __---3_ _- ' --- - ' - - - - ' - -- 0 - - ' - - --'--- - - ' - -- 500 Horizontally integrated smoke density kg m after one hour 20 __JL-- L - - - ' 1000 Particle size has a particularlv significant effect on radiative p operties_ To compare the effects of different particle size we have done calculations in which the total smoke mass is constant and the total number of particles varies with size i e manv small particles or fewer larger particles Spherical particles that are much smaller than the wavelength of sunligh t much less than 0 1 µmin diameter are effective absorbers but poor scatterers 0£ light Particles that are much larger than the wavelength of sunlight greater than 1 0 um in diameter are both poor absorbers and poor scatterers Particles that are similar in size to the wavelemrth of sunlight the center of the visible olar spectrum is near 0 5 um are very effective scatterers of light and may or may not be effective absorbers depending on their composition Because real smoke is made up of a distriburio of particle sizes it is necessary to perform numerical calculations to detennine the wavelength dependence of the absorption and scattering properties of the smoke For smoke fr urban fires the fraction of the total attenuation of light scattering plus absorption that is due to scattering ranges typically from 50 to 70% The railiative properties of smoke al depend on particle shape and whether the particles are homogenous or nonhomogenous in composition All of the climate assessments done to date have assumed that the smoke particles are homogenous spheres The microphysical processes of coagulation and condensation discussed below affect the composition size and shape o particles If the particles are mostly liqui or if liquid condenses on smoke particles then they tend to be spherical If on the other hand the particles are dry olid material then they tend to form long chains Soot in particular is noted for its tendency to form long chains Particles that are nonspherical absorb and sea er light more effectively than a sphere of the same volume For example an ellipsoid with a length three times i width is up to 20% more effective in scattering and absorbing light than a sphere of the same volume DEFE SE PRO GRAMS The u rface cloud and atmo phere a tempera ture near 27°C radiate long ·avelen th -l- to 20-µm infrared radiation The o-called greenhouse oa e in the normal atmo phere pnncipally water vapor and carbon dio ide ab orb and trap ome of the infrared radiation emitted from the urface and lower atmosphere and maintain the average earth temperature abo e freezing ee Re 13 A cloud of blacl- moke with particle in the rnJCrometre range would if sufficient den e ab orb most of the incident sunlight before it reaches the earth's urface The land surface would cool by radiating in frared radiation as it does on a clear night The same micrometre-size _moke particle that i a good ab orber in the ·1sible is a relatively poor absorber dm ·n by about a factor of ten in the infrared and therefore is not very effective in slowing the los of infrared radiation from the earth If however there is ubstantial coagulation of the moke particle and particularly if they form chains comparable in length to infrared wavelengths then absorption at rhe e ·avelengths can become more effecfa·e and the radiative cooling of the darkened earth would be slowed The radiative properties of smoke can ary greatly depending on its physical properties which are controlled in tum by microphysical and chemical processes The effects of variations in the optical propertie of smoke on the estimated climatic impact for large smoke loadings have been investigated by Ramaswamy and Kiehlw and by Luther 21 For a given mas of smoke in the atmosphere the e timated surface-temperature reduction ca n vary by about a factor of two for variations of the optical properties between plausible limits Microphysics In our research the term microphysics refers to the physical and chemical interactions of smoke particles ith one another and with the environment Two major microphysical processes coagulation and scavenging have the potential to affect substantially the projected climatic response to a ma ive smoke injection Coagulation involves smoke particles colliding and adhering to form larger particles It determine the size distribution of the assortment of smoke particles and is the primary factor in determining how much the particles affect olar and terrestrial radiation fluxe The e £luxes in tum determine urface temperatures and the climatic response Scavenging involves the interactions of smoke with water vapor condensed water ice and snow It can lead to the removal of smoke particles from the atmosphere by rain or snow and is the primary determinant of how long particles remain in the atmosphere either the study of Crutzen and Birks nor the TTAPS study included a detailed consideration of coagulation or cavenging during the spread of smoke at early times after injection Because o many microphysical interactions are possible coagulation can be important on both the short time scales that are associated with the lofting of smoke plumes above a fire and on the longer time scales that are associated with the spreading of smoke on a global cale Coagulation was considered in the TTAPS study but because it assumed that smoke was instantaneously spread throughout the northern hemisphere the particles were on average so far apart that they seldom collided and did not coagulate to a significant extent We have developed a new computer code that is capable of describing the coagulation of particles as they become dispersed At early times within dense smoke plumes the particles tend to be close together and to coagulate quite rapidly At later times but still in the plume phase the particles are so widely dispersed that coagulation can be ignored During the time of plume injection the first hour we find that unless the emission rate for smoke is much greater than 3% of the fuel being burned the coagulation of particles is not sufficient to alter significantly the optical properties of the smoke In contrast even with conservative estimates of the initial smoke density expected on regional scales our results shown in Fig 4 and 5 indicate that the particles may coagulate 21 enough in the first week to decrease their ability to absorb and scatter solar radiation by up to 40% Optical depth is the term used to measure the vertically integrated extent of absorption and scattering A 40% decrease in optical depth is equivalent to having 40% less smoke These effects have been investigated by Penner 22 The microph ysics of scavenging is even more complex than that of coagulation Interactions between smoke and the water in clouds rain or ice must all be taken into account Precipitation and the removal of smoke by scavenging may occur as a result of such interactions In the normal atmosphere a balance is reached between the number of cloud condensation nuclei or seed particles entering a cloud and the amount of available water vapor Under such conditions water vap or condenses on the nuclei forming droplets that further coalesce and grow into raindrops or snowflakes that can become large enough to fall to the ground Fig 4 The number of smoke particles per unit of radius is shown as a function of particle radius for an initial loading of 800 pg cm3 Loading is a measure of the density of smoke The size distribulion of smoke particles helps to determine their optical properties Particles that are in close proximity in the smoke higher on the curves coagulate to form fewe r larger particles Within seven days after an injection of smoke the number of particles in the optically important size range between 0 1 and 1 0 µm may be significantly reduced due to coagulation Fig 5 The extinction cross section of smoke is a measure of how effectively a gram of smoke interacts with light As the number of small smoke particles decreases over time see Fig 4 the extinction cross section also decreases The 40% decrease shown here reduces the optical depth of smoke by the same percentage and would be equivalent to a 40 % reduction in the amount of smoke if optical properties were not changing 22 1012 - ·2 1010 ' ' Cl Q 2 E 10 8 0 t ct' l ll Cl 106 1l o- E ' ' - 0 s 104 II Cl 0 102 E z 10° C -- 10 - 4 10 - 3 5 10 10 - 2 0 1 Radius µm 1 0 10 15 20 Time days 25 30 8 E g5 tiCl ' 1 l 4 e 0 g 2 0 w 0 0 Depending on their chemical composition smoke particles in the plume above a fire may act as cloud condensation nuclei However their number may be so great relative to the available water vapor that only very small water droplets can be formed Such droplets are too small to fall as rain at the same time they are too large to be moved enough by Brownian motion and other processes to coagulate into larger drops that would fall This condition of too many condensation nuclei is called overseeding and may substantially limit the scavenging of smoke particles by precipitation in a fire plume Crutzen and Birks mentioned the possibility that overseeding might occur but no attempt was made to assess its importance The TIAPS study noted but took no account of overseeding and instead simply assumed that 25% of the smoke would be scavenged through various mechanisms in the fire plumes If overseeding takes place rain is unlikely to occur directly above a fire Capping clouds which are observed above many fires form as the result of condensing moisture but we do not know whether the drops of moisture condense on the smoke particles generated by the fire or on the background aerosols that are naturally present in the atmosphere Indeed water vapor will not condense on all types of particles Small particles must contain a small fra ction of soluble material if they are to serve as condensation nuclei Hydrophobic or water-repellent material such as pure soot or oily particles will not fun ction as condensation nuclei wlless the particles are fairly large If on the other hand overseeding does not take place then it is quite possible that rain will occur above a fire From our calculations involving smoke plumes from intense fires we find that more than 1 5 times the amoun t of water that is usually necessary to initiate rain is available to condense In our model for an in tense fire the total amount of condensing water is 100 times by weight the total amount of smoke If rain fo rms falls and sweeps up smoke particles less smoke would be carried to high altitudes We do not have good DEFENSE PROGRAMS e timate of the extent of such cavenging It ha been suggested that the black rain ob erved from the fires after the bombing of Hiroshima was caused by uch cavenging Calculations with models that treat clouds b accounting for the dynamics of a fire and the rise of smoke plumes are comple use large amounts of computer time and do not yet treat the formation of rain in detail Although the models include equation to describe the formation of droplets ice and rain the adju table equation parameters are elected to yield results consistent with what happens in the normal atmosphere The models do not treat interactions with smoke or other pollutants The rain that might occur above a large-scale fire generated by a nuclear explosion requires a more detailed approach We are currently developing a computer model that is an extension of our coagulation model that will incorporate other relevant mechanisms and that will enable us to answer que tions regarding the fate of smoke Thi model will be run in conjunction with the plume model to assess earlytime scavenging and with the globalcale model to assess longer-term removal of smoke To assess the optical properties of make after it has come into contact with water we must model the interaction of smoke particles with water vapor and drop Some measurements have shown a significant change in the size distribution of smoke particles after interactions with a capping cloud above a forest fire Apparently interaction with the cloud either enhances the coagulation rate of dry smoke particles or causes them to swell as they accumulate water in either case creating more large particles Recall that spherical particles tend to become less efficient absorbers of sunlight as they grow larger than the wavelength of the light We do not know however how effectively light might interact with irregularly shaped swollen smoke particles It is believed however that a carbon smoke particle surrounded by water absorbs more visible light than the particle and water drop acting separately C limate Changes Due to Smoke Injection As we have cliscu sed the presence of smoke particles in the atmo phere perturbs the normal passage of solar radiation Initial studies by everal group using numerical models have shown that sub tantial surface cooling can occur if large amounts of highly absorbing smoke particles are injected high into the troposphere A major focus in presenting these results has been on developing estimates of the temperature change on land surfaces in the midlatitudes of the northern hemisphere following the injection of 150 to 250 Tg of smoke Some of the predicted global consequences of such an injection together with possible effects from injections of nitrogen oxides and radionuclides are illustrated in Fig 6 The TTAPS study used a globalaverage model of the atmosphere's vertical structure These authors estimated that under a hemispheric cloud of about 200 Tg of smoke land-surface temperatures in midcontinental regions would drop by from 30 to 40°C from an initial average temperature of about 13°C to temperatures well below freezing Temperature drops over the oceans were calculated to be small It was the predicted severe cooling over land that led to the coining of the term nuclear winter Subsequently calculations with more detailed models have indicated that substantial cooling may occur over land but may not be so drastic as the TTAPS estimate The TTAPS estimate of cooling by 40°C in midcontinental regions for example should be reduced by roughly one-half to account for the effect of ocean buffering that is the modifying effect of warmer ocean air when considering average temperature changes over land MacCracken 23 derived an early value of an 8°C hemispheric average temperature reduction over land using a two-di mensional climate model The midlatitude temperature changes on land under the smoke however were fo und to be about a factor of two greater than the hemispheric average value The results of Covey et al from the 23 Fig 6 When account is taken of the various assumptions made in early globa leffects studies an injection of about 150 lo 250 Tg of smoke into the troposphere would resu lt in cooling of land surfaces by from 25 to 35° C in summer in midcontinental regions of the north· ern hemisphere Many of the environmental effects that have been predicted by scientists at LLNL and elsewhere will be modified and refined as the physical uncertainties are further reduced ational Center for Atmospheric Research indicated a cooling of about 25° C in summer and of about 10°C in spring and a few degrees in winter during the fir t few weeks after in jection of smoke Their e timates may be ome •v hat too large becau e the diurnal cycle land-surface heat capacity and the effect of scattering solar radiation were neglected So iet estimate of the temperature reduction are somewhat larger but they in effect assumed a much larger moke injection than the other tudies 25 Thu early studies of the average midlatitude temperature reduction in the northern hemisphere range from about 15 to 25°C when account is taken of the varying a sumpt ions in the models used by these investigators Predicted midcontinental temperature drops are from 25 to 35° The expected biological and ecologica1 effects of this cooling which is more moderate than suggested by TfAPS would nonetheless remain significant Even this apparent agreement among studies is however somewhat deceiving in that all of the climate models included many additional assumptions and simplifications that have not yet been evaluated For example none of the models treated the movement of smoke from the scale of individual fires to the global scale Thus they did not account for the patchiness of smoke that would be created by the scattered sources Total solar radiation reduced by dust Total solar radiation reduced by smoke at northern-hemisphere midlatitudes Lowered tropopause Reduction of ozone layer Injection of 150 to 250 Tg of smoke Increase in ultraviolet flux at 0 30 µm · - -· Radionuclides Upper tropospheric warming by 30 lo 50°C - -- • - -·- · - · Average dose of 20 rem · - ·· • Increased land-sea temperature contrast • may increase storminess Lower tropospheric cooling Temperature reductions Cooling of midcontinent land surfaces may reach 25 to 35° C in the northern hemisphere for massive injections of smoke 24 buffered via heat supplied by oceans DEFENSE PROGRAMS changing weather and localized cavenging Furthermore none of the model even attempted to relate the remO ·aJ of moke to the actual scavenging proce e that are operative These relation hip will be e peciall difficult to model becau e mo t precipitation sy terns are smaller than can be easily re olved in global-climate models As a result me o caJe models which can treat cloud y terns with a resolution of everal ten of kilometre must be modified and used to investigate scavenging rates Finally many of the assumptions and representations of various processes in the climate model e g the effect of smoke on olar and terrestriaJ radiation are not yet accurate enough to ensure that changing a given formulation will not significantly aJter the model results Three-Dimensional GlobalScale Simulations Re earchers at the Laboratory have recently taken two preliminary steps to addre s some of the shortcomings in the early climate calculations The first invoh·es a eries of sensitivity studies to e 'aluate the importance of various assumptions and approximations that were made in early calculations The second invol es the development of significantly improved models that make fewe r assumptions Our approach has involved the coupling of two models The first is a two-vertical -layer three-dimensional general circulation model GCM of the atmosphere that was developed by Oregon State University OSU The econd is a three-dimensionaJ model of trace species and microphysics called GRA TOUR that is being developed at L L The models are coupled in a fuJly interactive manner 1n cooperation with scientists from the State University of New York at Stony Brook and from OSU we have conducted a series of model simulations to assess the sensitivity of the potentiaJ climate response to various assumptions We have tested the sensitivity of the results to assumptions concerning the amount of smoke injected the altitude distribution of the smoke the opticaJ properties of the particles and the manner by which the effect on solar radiation i calculated For these ensitivity studies we have assumed that smoke is spread evenly through most of the northern hemisphere and remains fixed as has been assumed in earlier calculations Our results indicate that climatic cooling is greater if the smoke is injected into the upper troposphere rather than into the lowest few kilometres The increa ed cooling occurs because absorption of solar radiation takes place above the level where greenhouse gases can trap and then radiate substantial amounts of that energy downward to the surface of the earth 13 1n previous calculations solar radiation has been treated in a highly simplified manner The caJculations have often assumed either that smoke does not scatter any solar radiation or that cattering can be accounted for by simply reducing the estimated amount of injected smoke ln addition most calculations have fixed the position of the sun holding it at an appropriate average angle above the horizon rather than allowing it to follow its normaJ daily cycle The first approximation is inaccurate because particles of the size with which we are concerned also tend to scatter solar radiation with most of it directed downward and toward the surface of the earth The ultimate effect of such scattering is to reduce the amount of cooling The second approximation is inaccurate because solar absorption varies exponentially rather than linearly with the amount of smoke along the path of the sun's rays Thus significant amounts of solar radiation can reach the surface near noontime through modest amounts of smoke aJthough little solar radiation reaches the surface when the sun is low on the horizon Such approximations are particularly important for quantities of smoke that are less than the massive amounts considered in the TTAPS article 1n another series of sensitivity studies we investigated the effects of varying the optical properties and amounts of smoke When the smoke is either less absorbing or the amount of injected smoke is reduced calculated temperature 25 Fig 7 Simulation of the expected temperature changes ten days after smoke injections of varying amounts into the northern hemisphere between about 20 and 70 deg north latitude The different values of optical depth r from top to bottom are approximately equivalent to smoke amounts of 20 40 60 120 and 240 Tg respectively The cool ing ettect becomes negligible for the case of r 1 5 which is equivalent to reducing the amount of injected smoke suggested by TTAPS by about 75% The smoke particles in this simulation absorb 30% of the solar radiation incident upon them The coo ling would be somewhat greater for particles that absorb 50% of the incident solar radiation Smoke is assumed to be distributed with equal mass-mixing ratios up to about 11 km reductions are dramatically decreased As shown in Fig 7 if smoke is reduced to about one-third the amount suggested by TTAPS the hemispheric-scale cooling effect essentially disappears Such changes in the amount of smoke that might be injected into the atmosphere are well within the current range of uncertainty GRANTOUR is a three-dimensional model that can transport smoke particles and estimate the mlcrophysical and scavenging processes that affect their concentration and distribution To treat these processes more accurately and efficiently the atmospheric formulation for GRANTOUR is somewhat different than that for our climate model GRA TOUR divides the troposphere into about 10 000 parcels of equal air mass spread evenly throughout the atmosphere from the surface to 11 km In the coupled GRANTOUR and GCM model the parcels are moved by the three-dimensional wind fields calculated by the GCM The air parcels also carry with them information about the concentration of smoke particles injected into the air mass Precipitation may occur depending on what is taking place in the nearest grid cell of the GCM In general the GCM portion of the simulation includes treatment of -5 - 15 - 20 - 25 L___ 0 _JL - ---'- --- '--- - - ' - - - - - ' - -----L-----1- - - - - 1 - - _ J 20 40 60 80 Latitude deg 26 tropospheric dynamics and thermodynamics and of the effect of soot on solar radiation The GRANTOUR part of the simulation includes treatment of particle transport mlcrophysics and scavenging by precipitation Although our interactive model coupling the GCM with GRANTOUR treats some new processes it still has many limitations Perhaps the most important problem in conducting globalscale simulations is the development of more accurate algorithms for representing the scavenging process This is a particularly difficult problem because actual precipitation occurs on scales that are smaller than the parcels treated by the climate model Furthermore there are no simple ways to verify that the approximation is working properly At present we relate the scavenging rate for smoke to the removal rate of water vapor in the atmosphere The particle lifetime based on rates of evaporation and precipitation is a little more than a week in the normal atmosphere We follow two classes of smoke particles those with diameters less than 1 µm and those with diameters of 1 µm and larger With our assumptions the larger smoke particles are removed a few times more rapidly than the water vapor because of the many processes that can remo e large particles in a cloud Smaller particles less than 1 µm in diameter are removed somewhat less rapidly than water vapor because such particles tend in ways not clearly understood by microphysidsts to avoid capture Testing and verification of these approximations is the subject of current research At present for these simulations we assume that the optical characteristics of smoke particles remain constant in time an assumption that may overestimate optical depth as particles coagulate Another imp01tant limitation of the simulation is that the resolution of the 10 000 parcels in the GRANTOUR model is only adequate to treat scales on the order of a few hundred kilometres on a side Finally we do not treat the possibility that the smoke could be heated and rise into the stratosphere our newest version will include this DEFENSE PROGRAMS Our most recent studies prescribe solar declination at its July value July was selected for our calculations because it appears to be one of the most sensitive month in terms of the potential climatic effects of smoke As expected the resulting predicted conditions over land for the normal climate unperturbed by smoke are somewhat warmer than would actual be observed during July because the model assumes that solar conditions for July have persisted indefinitely This extended period of July solar radiation also tends to dry out the continents and results in somewhat less precipitation than is actually observed For the period we are now studying the fir t month or so after the injection of moke exchange we do not allow the temperature of the ocean to change from its observed July values Although changes in ocean temperature are probably not important during the first few weeks they may become significant as we begin to study longer periods Thus despite our many improvements care must be taken in interpreting our results because the present necessary approximations and assumptions may affect the outcomes of the interactive model in ways that are not yet well understood At this stage results from models at LLNL and elsewhere should still be viewed not so much as predictions of what will happen but rather as sensitivity studies of what processes are important in determining the potential climatic perturbation of large amounts of smoke in the atmosphere Transport and Distribution of Smoke For the simulations we have run to date our basic reference calculation assumes an injection of 150 Tg of smoke into the troposphere 26 This amount of smoke is equivalent to the TTAPS 4 and Academy 6 estimates for urban fire emissions in their baseline scenarios We have neglected any smoke from forest fires TIAPS assumed 80 Tg and the Academy report assumed 30 Tg from this source As previously discussed recent studies suggest that the amount of such smoke is likely to be much less than the urban smoke that it is a poorer absorber than urban smoke that it would be injected at lower altitudes and that it is thus expected to affect the climate much less than urban smoke We refer to the 150-Tg simulation as an interactive calculation because the smoke is moved through the atmosphere by winds and can be scavenged by precipitation The smoke in tum alters the radiation distribution in the atmosphere and thus the calculated climatic conditions Our analyses have focused on the pattern of spreading smoke and on the resulting changes in surface temperatures Figure 8 shows the distribution of smoke on several days after a nuclear exchange from a vantage point centered approximately over the mid -Atlantic Ocean Each dot represents about 4000 tonnes of smoke spread throughout a volume of 550 000 km3• The smoke is injected at five locations the western U S the eastern U S Europe and two areas in the Soviet Union This distribution of sites does not come from a detailed war scenario but simply places the emissions where very large numbers of major fires seem plausible During the first day Fig 8a the smoke has not spread much beyond the source regions Thus we can immediately appreciate the substantial departure from previous models that assume an instantaneous and uniform distribution of smoke throughout the northern hemisphere 4•6•23•24 The maximum optical depths over Europe and Asia are as large as 50 at this time Somewhat more rapid ruspersion by stronger winds over orth America yields maximum optical depths of about 20 Although many water clouds have optical depths of ten or more light is still able to pass through because the cloud droplets scatter rather than absorb the solar radiation For smoke from urban fires once optical depths reach three to five virtually no solar raruation would reach the surface at average angles of solar elevation If this smoke were spread uniformly over the hemisphere the average optical depth would be about three roughly equivalent to the TTAPS estimate for a similar case 27 By the fifth day Fig 86 easterly ·winds in this meteorological simulation have carried smoke from western orth America toward the southwest Smoke from eastern orth America has been carried across the Atlantic Ocean and Eurasian smoke has swept across Asia toward the Pacific Ocean Maximum optical depths have decreased to below 20 however the average northern- Fig 8 GCM-GRANTOUR simulation of smoke distribution following an initial injection of 150 Tg of smoke Each dot in a represent about 4000 tonnes of smoke spread throughout a volume of 550 000 m about_450 by 450 by 3 km on the first day of an exchange Patchiness continues to be 28 evident on the fifth day b and the tenth day c By the twentieth day d smoke spreads more uniformly over the northern hemisphere except in low latitudes Some smoke also spreads to equatorial and subtropical latitudes of the southern hemisphere DEFENSE PROGRAMS hemi pheric optical depth remains about the same a on the fir t day because cavenging i quite slow for the small make particles that are important in determining optical depth Patchine s persists throughout the calculation and i quite evident even on the tenth da Fig c We may actually be undere tirnating patchlness because of our appro imation in calculating the cavenging process The GCM assumes that rain occurs uniformly over a relatively large area whereas in fact heavy rainfall tends to occur in more localized regions Thus greater patchiness may be expected after we refine the model to include more discrete regions of varying precipitation This is an important issue because smoke particles ab orb light exponentially The total hemispheric absorption of solar radiation and therefore the change in hemispheric average temperature would decrea e as patchiness increases By the twentieth day Fig 8d the smoke has spread more uniformly over the northern hemisphere except in low latitudes Maximum optical depths are reduced to ten The average optical depth over the northern hemisphere is about two indicating some scavenging of smoke and also some spread of the smoke to equatorial and subtropical latitudes of the southern hemisphere by wind changes induced by the moke Another way to view the results of the same 150-Tg interactive simulation is to display concentrations of smoke in three different layers of the atmosphere Figure 9 shows results for the tenth day a High altitude above 6 km SON 40N 40S b Middle altitude 3 to 6 km SON 40N 40S c Low altitude 0 to 3 km SON 40N Fig 9 40S 150W 100W sow 0 longitude SOE 100E 150E Vertical distribution of smoke on the tenth day after an injection of 150 Tg This coupled GCM-GRANTOUR cal culation is an interactive simulation for moving smoke The smoke is patchy and somewhat denser in the uppermost layer of the troposphere 29 The lowest layer shown in Fig 9c extends from the surface to an altitude of 3 km the intermediate layer Fig 9b extends from 3 to 6 km and the highest tropospheric layer Fig 9a extends from 6 to 11 km Again each dot represents a mass of about 4000 tonnes of smoke We have assumed that the smoke is initially injected about equally in terms of mass in each of the three layers By the thirtieth day Fig 10 concentrations of smoke are reduced most in the lowest layer of the atmosphere because of dispersion and scavenging by precipitation The smoke has also spread into the southern hemisphere particularly in the upper layers where winds are stronger Such smoke could affect land temperatures in the southern hemisphere Temperature Response The calculated pattern of emperature change over time for our assumed injection of 150 Tg of smo e is d · in Fig 11 For days 1 to 10 Fig Ila average temperature decreases on land surfaces beneath dense smo e are as large as 20°C for exampl e o er midcontinental Asia Regions experiencing little or no smoke over this period such as China and sou ero Asia experience virtually no temperature change For days 11 to 20 Fig 11 the average temperature decreases by as much as 35°C over Asia In contras the temperature change o er orth America is far less than in Asia because the win in this particular simulation have carried much of the smoke off the continen and replaced the smoke with dean Pacific air a High altitude above 6 km b Middle altitude 3 to 6 km SON 40N Cl Cl a 0 Ill J 405 Fig 10 Vertical distribution of smoke for the thirtieth day after an injection of 150 Tg In this coupled GCM-GRANTOUR interactive calculation concentrations of smoke are entering the upper atmosphere and are reduced more in the lower layer of the troposphere because of dispers ion and scavengi ng by precipitation 30 40N Cl Cl a j 0 405 Longitude 150E DEFENSE PROGRAMS a Days 1 to 10 90N 60N 30N Q O a ca 0 J 60$ 90S b Days 11 to 20 SON 60N 30N Q O ca 0 J 30S 60S 90S c Days 21 to 30 90N 60N 30N Cl O £ ni 0 J 30S 60S 905 180W 90W 0 Longitude 90E 180E Fig 11 Average changes in surface temperature over land for a 150-Tg injection of smoke from five source regions The smoke is interactively transported and scavenged by the climate model a Days 1 to 10 b days 11 to 20 and c days 21 to 30 31 Clearly the location of the smoke determines where the temperature changes will occur In addition ocean buffering the heating effect over land by air that has been warmed over the oceans is greater for orth America than for the larger continent of Asia From days 21 to 30 Fig llc the cooling begins to moderate in the northern hemisphere by a few degrees as some of th e smoke is carried to the southern hemisphere Only modest cooling occurs in the southern hemisphere for our July simulation cooling that is by no means as extreme as the quick freezes suggested by Covey et al i -1 We have also completed calculations for an injection of approximately 150 Tg of smoke assuming that it is immediately spread uniformly around the northern hemisphere and then held fixed in time and space The purpose of such calculations is to isolate the effect of moving smoke We refer to these simulations as uniform-smoke calculations As expected patchy smoke in the interactive case causes more extreme temperature reductions under regions of dense smoke and less change elsewhere than for the uniform-smoke calculations These differences are greatest during the first ten days After a few weeks the moving smoke spreads to cover almost uniformly the midlatitudes of the northern hemisphere and there is relatively little difference between our interactive and uniform-smoke calculations We have also made calculations assuming only 10% as much smoke is injected Once the smoke is spread uniformly injection of 15 Tg of smoke causes virtually no significant temperature changes in fact a little warming may even arise from a slightly reduced planetary albedo because the dark smoke is less reflective than either clouds or some parts of the earth's surface At early times however the dense patches can induce modest and localized cooling from several to tens of degrees Celsius The local changes are of course not as large as those following a 150-Tg injection To gain some perspective on the general character of predicted changes on 32 a day-to-day basis we have also plotted the predicted temperature variations over time at two particular locations Such numerical results should be viewed only in terms of their general features and not as quantitative estimates or projections Figure 12a shows temperature changes for a location in the central U S for a case with both interactive and uniformly spread smoke about 150 Tg and compares these two outcomes with our control simulation which has no smoke The effects are representative of a dry agricultural area in summer It is evident that the normal daily temperature cycle in the absence of smoke involves relatively large day-to-night variations In addition the high and low temperatures over the month are not constant rather cooler and warmer periods appear as different weather systems move across the midcontinental U S For the simulation assuming a uniformly spread smoke optical depth of three at this latitude less than about one-quarter of the normal light reaches the earth's surface The diurnal temperature cycle continues to be evident although the daily variation is not as large as that for the control simulation While the temperatures do decrease resulting temperatures are not substantially cooler than the normal cool periods in the control simulation without smoke Thus maximum daily temperatures with uniform smoke are not much cooler than minimum temperatures when no smoke is present While such changes may affect crops they are not immediately life-threatening Outcomes for a 150-Tg injection of moving smoke differ substantially from those for uniformly spread smoke The smoke is initially so thick over the central U S that virtually no sunlight reaches the surface Temperatures decrease rapidly and the diurnal temperature cycle disappears Following the period of sharp cooling temperatures recover to near normal for a short time because clean Pacific air replaces the smoky air mass Then as the smoke continues to spread other patches induce periods of cooling After thirty days when the smoke has spread more or less uniformly over the hemisphere surface DEFENSE PROGRAMS temperature reductions for the two simu lation with moke are similar The pattern of temperature variation for a location o er western A ia Fig 12b is imilar in character to that for North America However because A ia is a larger continent and the effects of ocean buffering of the temperature change are les effective the cooling is more evere Uniform smoke induces a gradual but teady temperature decrease to about 20 to 25°C below the control imulation For moving moke the diurnal cycle completely disappears indicating that virtually no sunlight is reaching the urface A very sharp drop in temperature occurs in the first few day and persists until the smoke spread more evenly over the hemisphere The control temperatures in our perpetual -July simulation for western Asia about 35 to 40°C are warmer than the ob erved temperatures during an average ummer the present calculations may be slightly underestimating the intensity and duration of subfreezing temperatures Future calculations which will incl ude easonal variations should more accurately treat this effect In general our interactive simulations predict relativel y large decreases in urface temperatures for several geographical regions Because these harp temperature drops start from summer rather than global-average conditions however we do not find indications of the extended deep freeze predicted by TTAPS The temperature changes may also be quite variable as smoke intensi ty varies Such variability makes assessment of potential biological and ecological consequences much more difficult than if a simple decrease to temperatures well below 0°C were to occur as the TTAPS report suggested We do not expect the potential atmospheric and environmental effects to become significantly less serious as model simulations become more complete unless the amount of injected smoke is reduced several-fold F uture Cooperative Research Many calculations now indicate that significant cooling would occur if massive amounts of smoke from burning urban areas were injected high into the atmosphere during the warm season The possibility of the cooling effect identified by Turco and his colleagues has now been confirmed by a number of researchers and is further supported in a review recently published by the ational Academy of Sciences 6 Considerably more effort is req uired however to understand whether substantial cooling is likely or is a remote possibility Recognizing the implications of this effect the D A and the Department of Energy have established research programs to investigate the problems Fig 12 Temperature changes over a thirty-day period for a the central U S and b western Asia The calculations for movi ng and un iformly spread smoke 150 Tg of injected smoke can be com pared with the control simulation no smoke The cooling effects of smoke are more severe in Asia than in the U S because Asia is a larger continent and the influence of ocean buffering is reduced For the interactive calculation with moving smoke the diurnal cycle that is apparent throughout the control simulation and the uniform-smoke case initially disappears and a sharp drop in temperatures occurs until the initial dense smoke cover thins as the smoke spreads out over the hemisphere a Central U S 320 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 300 i a f GI a E 280 260 b Western Asia 320 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 310 i300 GI a E 290 280 270 ' - - - - - - - - ' - - - - - - - ' - - - - - - - - ' - - - - - - ' - - - - - - ' - - - - - - ' 0 5 10 25 15 20 30 Time days 33 further In addition the ational Climate Program Office has coordinated development of a research plan that outlines additional necessary research 7 The major recommendations of this plan are for measurements of the characteristics of smoke emissions from fues of various sizes and greatly increased efforts to model atmospheric effects on the mesoscale and the global scale Another approach to advancing our understanding of the issues has been the convening of scientific meetings and workshops to discuss progress and to help identify and resolve uncertainties The Royal Swedish Academy of Sciences sponsored the 1982 study that led Crutzen and Birks to first quantify the smoke problem These authors have presented their results since then at numerous scientific meetings The TIAPS report was presented to scientists at a workshop in the spring of 1983 and to the public at a conference in the fall of the same year Soviet and LLNL researchers publicly presented their research results at an international seminar on nuclear war in Erice Italy fo r the first time in the summer of 1983 and again in 1984 We continue to discuss new findings as part of scientistexchange visits at Livermore Moscow and other scientific meetings An international study of the Environmen tal Consequences of a uclear War ENUWAR sponsored by the Scientific Committee on Problems of the Environment SCOPE of the International Council of Scientific Unions has involved a series of workshops held around the world A report summarizing committee findings is scheduled to be released in late 1985 Scientists are starting to prepare status reports on various aspects of their nuclear-winter research and are attempting to develop a scientific consensus that can be of use in informing both the public and their governments Although many uncertainties remain and scientists may alter their preliminary conclusions as a result of future research efforts the issue of the global environmental effects of nuclear war has focused renewed 34 attention on the need for all nations to strive for greater global stability and enduring peace U Key Words atmosphere-modeling stratosphere tropopause troposphere dimate-global indurei cooling computer code-general circulation model GCM GRA TOUR fallout fires-energy-release rate forest nuclear-generated urban nuclearexchange cenario war winter ozone radiationinfrared nuclear solar ultraviolet rainout make-coagulation condensation microphJ ics optical properties particles plume radiative properties scavenging spreading otes and References 1 Long-Term Worldwide Effects of Multiple udear-Weapon Detonations ational Academy of Sciences ational Academy Press Washington D C 1975 2 M C MacCracken and J S Chang Eds A Preliminary Study of the Potential Chemical and Climate Effects of Atmospheric Explosions Lawrence Livermore ational Laboratory Livermore California Rept UCRL-51653 1975 3 P J Crutzen and J W Birks The Atmosphere After a uclear War Twilight at oon Ambia 11 114 1982 4 R P Turco 0 B Toon T P Ackerman J B Pollack and C Sagan · uclear Wmter Global Consequences of Multiple uclear Explosions Sdence 222 4630 1283-1292 1983 5 P R Ehrlich et al Long-Term Biological Consequences of Nuclear War Science 21 2 4630 1293- 1300 1983 6 The Effects on the Atmosphere of a Major udear Exchange ational Academy of Sciences ational Academy Press Washington D C 1985 7 lnteragency Research Report for Assessing Climatic Effects of Nuclear War Report to the Office of Science Technology and Policy prepared by The ational Climate Program Office National Oceanic and Atmospheric Administra tion February 5 1985 8 C R Molenkamp Numerical Simulation of Self-Induced Rainout Using a Dynamic Con vective Cloud Model Lav 'Tence Livermore National Laboratory Livermore California Rep t UCRL-83583 1980 9 J B Knox Global Deposition of Radioactivity from a Large-Scale Exchange Intl Coni N uclear Wa 3rd Erice Italy 1983 also available as Lawrence Livermore National Laboratory Rept UCRL-89907 1983 10 Several recent studies have also projected that ozone red uction may result from chlorocarbon releases fro m aerosol spray cans and other sources For the relatively small but longerlasting ozone depletion brought about this DEFENSE PROGRAMS 11 12 11 14 15 16 17 way the health concern is with lhe long-term integrated ultra olet dose which has been related to some kinds of skin cancer J E Penner Tropospheric Response to a ' uclear hchange Intl Con1 uc War 3rd Ence llaly 19 3 also available as Lawrence Livermore ational Laboratory Rept UCRL9956 19 3 H Stommel and E Stommel The Year 'ithout a Summer Sci Amer 240 176 1979 The energy fluxes to and from the earth's urface atmosphere and space are discussed in Energy and Technology Review UCRL52000- -9 September 1984 p 11 R D Small and B Bu h Smoke Production from Multiple udear Explosions in S on-Urban Areas Science 19 5 in pres Resume of Discussions and Conclusions at the SCOPE-E U 'AR Workshop Paris 22-2-1 October 19 Scientific Committee on Problems of the Environment 1984 See for example G E Carrier F E Fendell and P 5 Feldman ''Firestorms Fire D rnam Heat Transfer 25 55-64 1983 These authors modified a semianalytic approach developed to describe the dynamics of small-scale fires by B R ilorton G Taylor and J S Turner 'Turbulent Gravitational Convection from laintained and In tantaneous Sources Proc Ro_1 Soc London Series A 234 1-23 1956 To match the predictions reported from the Hamburg fuestorm Carrier et al assumed that the mixing of hot plume air and cool background air was restricted far more than i actually observed in less intense fires P C Manins Cloud Heights and tratospheric Injections Resulting from a Thermonuclear War CSIRO Division of Atmospheric Research Mordialloc Australia 1984 Manins used the theory of Morton et al noted in Re 16 directly but treated fires a point sources 18 19 20 21 22 23 24 25 26 J E Penner L C Haselman and L L Edwards Buoyant Plume Calculations A AA Aerospace Sci Meeting 23rd Reno evada 1985 also available as Lawrence Livermore ational Laboratory Rept UCRL90915 1985 W R Cotton Atmospheric Convection and uclear Winter Amer Sd 73 1985 in press V Ramaswamy and J T Kiehl Sensitivities of the Radiative Forcing Due to Large Loadings of Smoke and Dust f Geophys Res 90 19 in press F Luther Uncertainties of Radiative Properties of Smoke and Their Effect on Oimate Assessments Proc Intl Sem uclear War 4th Erice Jtaly 1984 E Penner Smoke Inputs to Climate Models Optical Properties and Height Distribution for udear Winter Studies Proc Intl Sem udear War 4th Erice Italy 1984 M C MacCracken uclear War Preliminary Estimates of the Climatic Effects of a Nuclea r Exchange Proc Intl Sem udear IM r 3rd Erice Italy 1983 C Covey S H Schneider and S L Thompson Global Atmospheric Effects of Massive Smoke Injections from a udear War Results from General Circulation Model Simulations ature 308 21-25 1984 V V Aleksandrov and G L Stenchikov On the Modeling of the Climatic Consequences of uclear War in Proc on Applied Mathematics of the USSR Academy of Sdence Moscow 1983 V V Aleksandrov and G L Stenchikov On a Computational Experiment Modeling the Climatic Consequences of uclear War '' Computat Math and Math Phys 14 No 1 140 1984 in Russian M C MacCracken and J J Walton The Effects of Interactive Transport and Scavenging of Smoke on the Calculated Temperature Change Resulting from Large Amounts of Smoke Lawrence Livermore ational Laboratory Rept UCRL-91446 1984 35