Committee on Impacts of Stratospheric Change
Assembly of Mathematical and Physical Sciences
Committee on Alternatives for the Reduction of Chlorofluorocarbon Emissions
Commission on Sociotechnical Systems
The deleterious effect of UV-B on human beings is related to prolonged, deliberate, as well as inadvertent sunlight exposure, which leads to the development of nonmelanoma skin cancer and malignant melanoma of the skin.
The epidemiologic evidence for the causative role of damaging ultraviolet radiation (DUV) in the development of nonmelanoma skin cancer of the exposed areas (the face and neck) is indisputable and is based on (1) latitude dependence and (2) high incidence in a susceptible population. The susceptible population includes white, fairskinned persons who sunburn easily and who receive prolonged occupational or recreational exposure. Further more, preliminary data suggest that there is a subset of this white population who have been shown to have a deficient DNA repair mechanism.
Although the causative role of DUV in the pathogenesis of malignant melanoma of the skin is less firmly established, recent evidence supports the previous conclusions of CISC that the development of malignant melanoma of the skin in man is related to DUV exposure. Previous evidence established an inverse relation of latitude and melanoma incidence, a low incidence in more pigmented peoples, and an increased number of primary melanomas on the habitually exposed areas (face) and relatively exposed areas (trunk and legs of females), with infrequent numbers of primary melanomas on the rarely exposed areas. More recent evidence substantiates, with a larger number of patients (761), this same body distribution.
It is now known that there is a strong cohort effect (persons who are age 30 now have a higher incidence of melanoma than persons the same age born 10 years earlier). This would suggest either a new causative agent (virus or chemical) or, more likely, a change in behavioral pattern regarding recreational exposure (increased leisure time, less concealment, more vacations in sunny areas).
Some current studies show that age-specific incidences are higher in persons who have resided longer in sunny areas. Those persons born in Israel had a higher incidence than persons residing in Israel but born in northern Europe.
There is a well-documented, recently reported dramatic increase in the incidence of primary melanoma in Norway and Sweden, presumably related to increased recreational exposure.
Recent analysis of British occupational mortality data shows that, for both employed males and their wives, there is a strong relationship between mortality and the socioeconomic class, with the highest incidence in teachers, businessmen, professionals, and higher-level administrators.
Expected increases in melanoma mortality based on current cohort data, without regard to changes in the ozone layer, show a rise from the current (1971-1975) rate of 26.3 per million for U.S. whites to 28.7 in 1976-1980 and 33.5 in 1981-1985. The effect of ozone depletion and increased UV-B would be added to this expected incidence.
GENERAL ASPECTS OF SUNLIGHT EXPOSURE
A few decades ago, the western world began to believe that exposure to the sun promotes health and the more exposure, the better. The dangers of general exposure came more clearly into view as sun-seeking habits became more and more widespread. Unfavorable effects range from transient sunburn through wrinkling and keratoses to skin cancer. We shall briefly describe these different effects below.
A single 15- to 20-min exposure of white skin to the sun in middle latitudes during summer can produce sunburn. This is an exposure to about 1.2 x 10 J/m. In perhaps more than two thirds of the white population, careful, repeated exposures will develop a protective tan. In the remainder, little or no protective tanning occurs, and the sensitivity continues, leading to repeated sunburn on exposure.
The longer-term effects of exposure to the sun involve the accumulated dose of exposure, usually over a period of years. Most wrinkling of the skin of the face and hands and all the warty thickenings of the skin called solar keratoses are associated with prolonged or intensive exposure to sunlight. While these effects are far from desirable, their seriousness is limited.
Against unfavorable effects of UV-B, ranging from sunburn to skin cancer, we must place in the balance one favorable effect of widespread importance:
This action is important to all of us in the world but fortunately does not require large doses of UV-B.
Attention is often directed to sunlight-caused cataracts. Animal studies show that the action spectrum for single-exposure induction of lenticular cataract peaks at 290 to 370 nm, and the threshold dose is above the dose that causes symptomatic photokeratitis.[1,2] There is concern that prolonged and frequent exposures may cause gradual cumulative photochemical alterations in the lens leading to cataracts in the absence of corneal changes. The action spectrum for this effect may be at longer wavelengths, and tryptophan may act as one chromophore. Epidemiologic data[4-7] suggest that solar ultraviolet may be causally related to some forms of senile cataracts, but much more information is needed to know the size and action spectrum of this threat. Ultraviolet-radiation effects on the lens have been summarized recently.
1. B. S. Yamanashi, H. Hacker, and G. K. Klintworth. 1979. Wavelength dependence and kinetics of UV-induced free radical formation in the human cornea and lens. Photochem. Photobiol. 30:391-395.
2. D. G. Pitts. 1974. The human ultraviolet action spectrum. Am. J. Optom. Physiol. Opt. 51:946-960.
3. S. Zigman, T. Yulo, and J. B. Schultz. 1974. Cataract induction in mice exposed to near UV light. Ophthalmol. Res. 6:259-270.
4. S. Duke-Elder. 1972. System of Ophthalmology, Vol. 14: Injuries, Part 2: Nonmechanical Injuries. C. V. Mosby, St. Louis.
5. R. van Heyningen. 1975. What happens to the human lens in cataract. Sci. Am. 233:70-81.
6. R. Hiller, L. Giacometti, and K. Yuen. 1977. Sunlight and cataract: An epidemiological investigation. Am. J. Epidemiol.105:450-459.
7. S. Zigman, T. Yulo, T. Paxhia, S. Salceda, and M. Datiles. 1977. Comparative studies of human cataracts. Abstract of the Association for Research in Vision and Ophthalmology, Sarasota, Florida.
8. J. A. Parrish, R. R. Anderson, F. Urbach, and D. Pitts. 1978. UV-A: Biological Effects of Ultraviolet Radiation with Emphasis on Human Responses to Longwave Ultraviolet. Plenum Press, New York.
SKIN CANCER CAUSED BY EXPOSURE TO SUNLIGHT
The serious effects of exposure to the sun involve the development in some individuals of skin cancers of three major kinds:
Together the first two types of skin cancer add up to the most frequently detected cancer in man and have an increased incidence over the past decade (J. Scotto, NCI, personal communication).* They are also the most easily and most successfully treated human cancers. The quantitative extent to which agents other than UV exposure cause nonmelanoma skin cancer in the white population has not been established. It is, however, believed to be small. Some nonmelanoma skin cancer is caused by exposure to arsenic, pitch, and x rays, often in the course of work, sometimes following treatment of skin disorders. This latter group of tumors is found among patients of all degrees of skin pigmentation who happen to be exposed to these agents. In some less-developed countries, most nonmelanomas seem to arise in neglected wounds (Camien et al., 1972; Fleming, 1975).
The report of the Climatic Impact Committee (1975, page 40) stated that "The available evidence indicates that the spectral sensitivity for skin cancer is similar to either the action spectrum for erythema production or the spectrum for damaging DNA. The two spectra are similar but not quite identical." (See Figure 3.1.) Nothing has been learned since to weaken this conclusion.
Nonmelanoma skin cancer (which rarely leads to death) is at present a serious problem because of disfigurement (frequently minor and infrequently severe) and the significant economic burdens associated with its treatment.
Melanomas, the third mentioned cancer, are a serious life-threatening hazard and are more common than primary malignant brain tumors and twice as common as Hodgkin's lymphoma (Silverberg, 1979). The most recent figures, for the years 1965-1969 (Cutler et al., 1975), show that only two thirds of new melanoma patients survived for five years, about the same fraction as for breast cancer.
The Climatic Impact Committee Report, 1975, page 41, stated that "Although the evidence associating UV-B with malignant melanomas is not so strong as for nonmelanoma, we believe that the only action spectra that we can prudently use for any quantitative estimate of the potential hazard arising from an increase in UV-B are those given in Figure 4 [repeated here as Figure 3.1]." The evidence collected since that report serves only to strengthen that conclusion, as we shall see below. Nevertheless, in developed countries, a fraction of malignant melanomas must come from other causes than ultraviolet radiation.
In terms of human health, then, our concern with the consequences of exposure to the sun
Skin cancers associated with solar-ultraviolet (UV) radiation, as diseases of the less-pigmented peoples, are thus a major concern in the United States, Europe, and Australia, and to emigrants from these regions to other parts of the world. (The as yet uncertain effects on plants, animals, and climate will be the main consequences of CFC release for the more pigmented peoples.)
GENERAL MECHANISM OF SOLAR SKIN-CANCER CAUSATION
It is generally accepted that changes in individual molecules of the body's DNA (deoxyribonucleic acid) are most often the initial key step in cell mutagenesis and cancer production. Accordingly, we expect that sunlight-induced cancers probably arise by the action of sunlight on DNA or compounds of analogous sensitivity to different wavelengths. This hypothesis is strongly supported by the increased skin cancer, including melanoma, observed in people who have DNA repair defects (xeroderma pigmentosum). Essentially no UV radiation of wavelengths less than 290 nm currently reaches the ground, and UV radiation of wavelengths above 320 nm (UV-A--320 to 400 nm) does not attack DNA seriously. Accordingly, whenever sunlight stimulates skin cancer, DUV (defined as UV-B in the wavelength range 290-320 nm, weighted in accordance with its effectiveness in altering DNA) should be the portion of the UV that is effective. UV-B is generally believed to be the action spectrum for nonmelanoma; this has been confirmed in animals. Because the divisions between UV-C, UV-B, and UV-A are neither phenomenologically exact nor agreed on, for critical work one should define UV radiation in more rigorous spectroradiometric terms.
EVIDENCE FOR SUN-INDUCED SKIN CANCER
Beyond these general considerations, evidence for sunlight as a major or contributing cause of skin cancers takes five forms (see Appendix F):
What patterns ought we to see in the first four of these forms of evidence if
Where total accumulated dose matters:
These are, of course, the characteristics shown by nonmelanoma skin cancers (see Climatic Impact Committee, 1975, pages 36-40).
In the second case, where dependence on exposure is more complex:
We shall see, in the remainder of this chapter, that what limited information we have about melanoma incident and mortality is quite consistent with the second case, where dependence on exposure is more complex and other contributing factors may be involved.
With the foregoing factors in mind, what can we say about the overall strength of the evidence implicating solar UV radiation in the causation of skin cancer and thus leading us to expect increased skin cancer when the amount of DUV reaching the ground increases?
For nonmelanomas, three kinds of evidence--latitude dependence (Figure 3.3), body location (Figure 3.4), and occupational differences--all combine to point closely to exposure to the sun as a prime cause and to increased incidence as a quite certain consequence of increased DUV.
The situation for melanomas is somewhat different. The latitude dependence appears well established (Figure 3.3). While melanomas are not heavily concentrated on the most-exposed regions of the body, they do appear to avoid the least-exposed regions, and differences in patterns of locations between sexes correspond to the differences in exposure to sunlight, for example, more melanomas on the legs of females and more melanomas on the scalps and ears of males (see Appendix G, Figure G.1). Occupational differences are fairly large, and, while they do not correspond to differences in total exposure they do seem to correspond to differences in patterns of exposure. Changes in mortality and incidence with time are large and can reasonably be associated with changes in exposure behavior.
This leaves us with a qualitatively well-established relation of solar UV radiation to nonmelanomas and a well-founded anticipation of increased nonmelanoma incidence if the DUV reaching the ground were to increase. The contribution of solar-UV radiation to the production of nonmelanoma skin cancers is thus a well-established health hazard of some magnitude and ought to be responded to accordingly.
Less firmly, but we believe persuasively, there is a large likelihood that solar-UV radiation contributes to the induction and/or development of melanomas and that an increase in the DUV reaching the ground will induce an increase in melanoma incidence and mortality. (Such an increase may be small compared with the increases now going on, presumably as a result of changes in patterns of exposure.)
If melanoma is induced by UV light, then increased exposure would be expected to result in earlier (on average) appearance of tumors. Just such a situation has been occurring with melanoma especially of the superficial spreading type for which a falling mean age of occurrence is being observed. This corresponds to higher rates found for younger individuals born in successively more recent decades (cohort effect).
If, as we believe, DUV plays an important role in many melanomas, the dose-response relation is complicated and is likely to involve many other variables.
To summarize, we believe that the relation of solar-UV radiation to melanoma ought to be taken as a likely health hazard of significant size and responded to accordingly.
The dependence of both incidence and mortality on latitude has been clearly documented for melanoma skin cancer, as had dependence of incidence on latitude for nonmelanoma skin cancer (Climatic Impact Committee, 1975, pages 36-41, from which Figure 3.3 is reproduced). Higher exposure to DUV is reflected in more skin cancer.
Worldwide plots of latitude data support a latitude effect on both incidence and death rate, but some deviations exist in certain countries and at some locations within countries. These deviations may result from the effects of microclimate, difference in fraction of susceptible population, or differences in completeness of registry data collection. These factors make comparisons between countries difficult.
As we move toward the equator, the sun is more nearly overhead, and the total amount of sunlight increases at all wavelengths. This increase is significant for solar UV-B, whereas UV-A and visible light vary to a lesser degree. In the UV-B region, the controlling factor is absorption by ozone, which increases sharply toward the poles, both as we move under higher ozone amounts and as the sun departs from the zenith. Thus, the accumulated dose of UV-B--or, more particularly, of DUV--is likely to be greater for those living in lower latitudes.
The effect of more sun may well be partly compensated for by changes in living habits that lead to a smaller fraction of time spent in the strongest open sun or somewhat enhanced by a life style that increases the period of time spent in sunlight exposure. Measures of the extent of this compensating (or enhancing) effect do not seem to exist.
Nonmelanoma Skin Cancer Most nonmelanoma skin cancers occur
While, as noted above, a small fraction of nonmelanoma skin cancers are clearly due to other causes, what we know about nonmelanoma skin cancer suggests that the accumulated DUV dose built up over years is the crucial factor.
Melanoma Skin Cancer Until recently, the relation of sunlight exposure to the development of melanoma has been difficult to interpret because the pattern of melanomas over the different parts of the body is different from the pattern of nonmelanoma skin cancer.
Incidence rates per unit area of skin, from Queensland, show significant excess melanomas on the generally or occasionally exposed sites of face, leg, neck, and arm in women and face, ear, neck, and back in men. They also show proportionally fewer tumors in parts of the body that are virtually never exposed (Elwood and Lee, 1975).
According to U.S. population-based incidence figures (Third National Cancer Survey, 1975), U.S. white men also have most melanomas on the face and head and trunk. Women have rates similar to men on the face and head and trunk but have higher rates on arm and leg. Limitation of available data, for example, to face and head combined, limits the extent to which these observations can be interpreted.
There are clear variations in incidence as latitude changes within the United States. Melanomas on totally exposed sites show the most, and those on rarely exposed sites the least, latitude dependence (Scotto et al ., 1976b). Melanomas of the partially exposed (relatively exposed) sites are increasing in incidence over time at a faster rate than those on sites that have always been totally exposed (Lee and Strickland, 1976).
Most melanomas seen in the Malignant Melanoma Group Cooperative Study involving four U.S. clinical centers (761 cases to date) occur on lightly covered or occasionally uncovered regions of the body (Sober et al., in press). (See Appendix G, Figure G.1.) A similar pattern of location of primary melanoma has been reported by Braun Falco (1974). Very few have been seen (either in males or females) on the regions ordinarily covered by bathing suits. The difference between sites of occurrence on males and females is believed by these investigators to reflect different covering practices. Thus, women who usually wear dresses have many more melanomas on the potentially exposed part of the legs than do men who usually wear trousers. Men, who expose their trunks to the sun during recreation, have more melanomas on their chests than do women. It is highly probable that the few melanomas seen in areas virtually always covered by relatively heavy clothing are due to causes other than UV light.
Part of the argument about the role of UV radiation (presumably UV-B) in causing malignant melanoma has been the idea that basal- and squamous-cell tumors of the skin occur in exposed areas, while melanomas occur also in skin areas covered by clothing. To evaluate the UV penetration of different fabrics, Infante (R. Infante and F. Daniels, Jr., Cornell University, manuscript in preparation) measured the spectral transmittance of a number of different fabrics, some with a series of different moisture contents. A recording spectrophotometer with an integrating sphere was used so that forward-scattered and fluorescent light are included in the measurements.
At 300 nm a white handkerchief showed 32 percent transmittance in a single layer, 15 percent in a double layer, about 8 percent in three layers, and about 5 percent, 3 percent, and 2 percent transmittance with four, five, and respectively. A nurse's white nylon uniform transmitted about 4 percent at 300 nm, a white dress shirt about 21 percent, a white T-shirt about 21 percent, and a white sheet about 12 percent.
Women's modern nylon stockings transmit about 70 percent over a wide range of wavelengths since the transmittance measurement is essentially a measurement of an open network. For contrast, Infante obtained some silk stockings ca. 1904 from a costume agency and found in the calf region around 30 percent transmittance at 300 nm and in the thigh region about 10 percent transmittance. Thus UV effects could be expected to occur on regions covered by present-day clothing, especially the upper trunk.
There is a clinical impression that a majority of cases of melanoma are seen in younger middle- and upper-class males and females who pursue active outdoor recreational activities but spend their working hours indoors. A direct correlation has been demonstrated between death rate and socioeconomic level. As income and degree of education rise, melanoma death rate rises (J. A. H. Lee, University of Washington, preliminary data).
In current urbanized U.S. society, a general picture of most melanomas occurring on lightly clothed or "on-and-off" clothed body areas on persons not in outdoor occupations is quite compatible with some mechanism associated with the DNA sensitivity and with the observed latitude dependence of melanomas on these particular sites.
Magnus (1977) of Oslo reported on the analysis of 2541 melanoma patients observed in Norway during the period 1955-1974. The analysis supports an exogenic cause of melanoma in this series; Magnus concludes that exposure to sunlight is the important factor. Both the tripling of the incidence during this period and a steep increase during adolescence are attributed to increased recreational sunlight exposure as a result of (1) increasing leisure coincident with economic progress, industrialization, and urbanization and (2) change in clothing habits with a trend away from total concealment. Northern Norway (72deg. N latitude) has only one half the incidence rate of southern Norway (58deg. N latitude). This is attributed to greater exposure associated with the warmer climate and a higher DUV intensity in southern Norway. As in other reported series, there were striking differences in the anatomical site of the primary cancer in males and females--males showing higher incidence in the neck and trunk sites, and females showing a three times higher incidence on the lower limbs. It is to be noted that there was no change in incidence on face or feet during the entire observation period (1955-1974), supporting the hypothesis that change in behavioral pattern is an important factor in the striking increase in incidence of malignant melanoma since World War II.
Large-scale analyses of mortality from malignant melanoma were made by the Registrar General in England and Wales for the years 1949 to 1953 and 1959 to 1963 (Lee and Strickland, 1976). As might have been expected (because of higher total exposures to the sun), the rate of nonmelanoma skin cancer was higher among unskilled and skilled than among professional and managerial-type workers. It was quite the opposite with malignant melanomas, for which the higher mortality was concentrated in younger professional and managerial workers, confirming the clinical impression mentioned above and quite consistent with a high efficacy of recreational exposure. The most recent British incidence figures confirm higher rates of melanoma in the middle class (Registrar General's Statistical Review, 1975).
For both employed males and their wives, there is a strong relationship between mortality* and socioeconomic class, defined in terms of occupation (Figures 3.5 and 3.6). Number V on the abscissa scales refers to unskilled workers, IV to semiskilled (including miners), IIIM to skilled manual workers (skilled craftsmen of all types), IIIN to skilled nonmanual workers, including clerks and salesmen, II to teachers and businessmen, for example, and I to professionals such as physicians, lawyers, and administrators. The classification lumps together some rather disparate groups, but in practice it is useful. The mortality gradient indicated in the figures is particularly interesting, because the more sophisticated groups are most likely to get better medical services, i.e., they would be expected to seek treatment with smaller and early melanomas. Outdoor activity on a professional basis does little to disturb the controlling effect of prosperity or lack of it on the melanoma mortality (Table 3.1).
Changes with Time, Plausibly Related to Changes in Exposure Behavior
Some estimate of the marked impact of changes in lifestyle leading to increased exposure can be derived from careful studies of the experiences of young white women before and after the Second World War, a time when widespread changes in dress and behavior began to occur. From 1911 to 1940, approximately four out of each million United Kingdom women aged 15-44 died each year from skin cancer. Since at these ages deaths from squamous-cell cancers are virtually unknown, and skin carcinomas are confined to males, these deaths can be assigned confidently to melanoma. In other words, in a country with limited sunlight, there was a background rate of almost four melanoma deaths per million among young women who did not deliberately expose themselves to sunlight.
In Connecticut, women of the same ages had a melanoma incidence rate of 12 per million between 1935 and 1939. Given low survival, this incidence rate compared well with the pre-1940 U.K. mortality. In both countries, there was a rapid post-1940 increase: to a mortality rate of 7.8 in the United Kingdom in 1966-1970 and an incidence rate of 48 in Connecticut in 1970-1972. The difference between twofold and fourfold increases, in the United Kingdom and Connecticut (see Table 3.2) is consistent with the different increases of sun exposure associated with different lifestyles.
Table 3.3 summarizes annual increases in melanoma incidence and mortality in different countries and compares two periods of observation, 1941 to 1960 and 1965 to 1970. Figure 3.7 depicts the trends in skin melanoma among whites in the United States.
Are Melanoma Incidence and Mortality Reasonably Consistent with a Major Role for Solar-Ultraviolet Radiation?
Earlier we set down the general outlines of the pattern of evidence to be expected if total UV dose were not the sole critical factor in solar-related cancers, i.e., UV-B might be a contributing but not sole cause, and pointed out that melanoma behavior fitted that pattern. We have now reviewed the details of the evidence concerning melanoma and ought to inquire further as to the reasonableness of a major role for solar-UV radiation in melanoma.
Ozone depletion will result not only in an increase in UV-B flux but also in a shift to shorter wavelengths that reach the earth. If the action spectrum of human skin cancers is similar to the action spectrum for DNA damage, a nonlinear and disproportionately higher increase of these tumors would be anticipated. That is, a given percentage increase in solar UV-B should yield a higher percentage increase in melanoma. If the action spectrum for either form of human skin cancer is not sensitive in the shorter wavelengths, then a lesser effect of ozone depletion might occur.
We do not intend to propose specific mechanisms--to do this would be inappropriate at this time. We do try to point out some classes of mechanisms that would allow varying degrees of covering (sometimes total, sometimes none or partial, the latter from semitransparent clothing) of some part of the body or intermittent recreational exposures to increase the incidence of melanoma.
We know that persons of intermediate sensitivity to sunburn develop sunburn after initial exposures, but eventually tan. On a generally exposed portion of the body, or when occupational exposure occurs daily throughout the year, the annual re-enhancement of protective tan would often take place in the spring when total daily doses are low. Thus no sunburn would be produced by the heavy doses of summer DUV--and it is possible that any melanoma-inducing effect of these doses would be greatly reduced. On a partially exposed portion of the body, or when recreational exposures are both intermittent and seasonably concentrated, the annual re-enhancement of tan would be likely to wait until summer, would result from much more intensive doses, and would involve considerable sunburn. Under such conditions, any melanoma-inducing effect would probably be stronger, although the total accumulated dose is less. Similar behavior could arise when clothing is sometimes completely UV-absorbing and sometimes not.
Experience with persons who sunburn and actual measurements of transmission of UV-B through clothing show that significant amounts of UV radiation do penetrate clothing. Other possibilities for development of melanoma underneath clothing would include combinations of UV exposure with other unknown factors, such as the presence of a susceptible melanocytic target or even mechanical friction. If photorepair occurs in man (it has not been demonstrated in mammals), it might provide an additional explanation for more response on incompletely covered regions.
What about the effects of changes in habits? What if people, particularly those most likely to be affected, those who sunburn easily and never tan, were to expose themselves less to the sun? How would we expect such changes to be reflected in melanoma rates and incidences? Suppose, for example, that total exposure (or total exposure in a given pattern) is the critical factor, and that incidence is proportional to total dose. Then, if all susceptibles reduce their accumulated exposure time (allowing for hour-to-hour and season-to-season changes in DUV) by 20 percent, thus receiving 80 percent of what might have been the dose, and if, at the same time, the ground-level DUV increases by perhaps 25 percent, the two changes could compensate, and the incidence of, and mortality from, melanomas would be the same. This might at first seem to say that the habit change took care of the increased DUV. Much more realistic, however, is the comment that if we could have kept the habit change, but not increased the DUV, we would have had perhaps 20 percent fewer melanomas.
Desirable habit changes, especially by susceptibles, can be very worthwhile, affecting general levels of incidence and mortality. But the losses from more DUV at ground level are likely to involve increases in skin cancer, by a fixed fraction, whatever base level we may be able to reach by habit change.
FORECASTING THE EFFECTS OF DUV INCREASE
If UV-B, which varies with latitude, is responsible for most melanomas diagnosed in the United States, the observed relationship between UV-B and the incidence or mortality of melanoma in the United States can be used to forecast the effects of change in DUV intensity on both melanoma incidence and mortality.
Statistical studies by the National Research Council's Panel to Review Statistics on Skin Cancer were given in the report of the Climatic Impact Committee (1975, pages 177-221). More recent studies include those by Scotto et al. (1976a) and Green et al. (1976). These studies have a number of characteristics in common, namely:
Scotto et al. (1976a) and Green et al. (1976) differ in their chosen dose-response relations, the one using logarithmic rates proportional to dose and the other rates proportional to dose to a power. Neither study makes allowance for varying behavior of cohorts, which may or may not be important to this aspect of their studies. Both studies, if taken at face value, would give percentage increases in nonmelanomas up to twice the percentage increase in DUV. These statistical studies serve as a basis for estimating the increased effect that results from more DUV but not as an aid in establishing cause and effect.
As noted earlier, it is our general understanding of agent-caused cancers and the direction rather than the amount or the detailed behavior of changes in incidence and mortality (as we move from poles to equator) that points to UV-B exposure as a cause of skin cancer. To say this does not deny the probable importance of other factors, acting either alone or in conjunction with DUV. The localization of melanoma in lightly or irregularly covered portions of the body may illustrate the importance of at least one factor that combines with DUV. Differences in melanoma mortality between places with similar accumulated DUV doses show the importance of other factors, some of which may act independently of DUV, although others seem likely to act concurrently with it.
Given our general information about human behavior as it depends on latitude, some will anticipate that the residents of lower latitudes accept a smaller fraction of the available DUV dose than do the residents in higher latitudes. Because of the greater amount of DUV available in lower latitudes, lower-latitude residents would otherwise receive much greater total doses. The difference in melanoma mortality--or in either nonmelanoma or melanoma incidence--between the two latitudes would then be less than that corresponding to the difference in the DUV availability, since behavior differences will partially compensate for the enhanced amount in lower latitudes. Others would expect residents in lower latitudes to accept a higher fraction of the available DUV, thus enhancing, rather than partially compensating for, the latitude difference in DUV intensity. There will also be some compensation owing to differences in the proportion of residents of higher sensitivity, reflecting ethnic composition.
What if the DUV reaching the ground increases, so that current intensities at higher latitudes now equal previous intensities at the lower latitudes? The accumulated intensities reaching the ground at the higher latitudes will build up, somewhat slowly, toward those previously occurring at the lower latitudes. The behavior patterns may also change slowly, but probably not to match previous patterns at the lower latitudes, since if only DUV is increased, the only cue for greater avoidance to the sun will be unchanged. The change in behavior will probably be slower than the change in accumulated intensity. Any changes in ethnic composition will be, relatively, still smaller and still slower.
As a result, the accumulated doses for individuals at the higher latitudes will rise toward values that may be slightly above those originally found at the lower latitudes. This will occur if behavioral compensation and ethnic composition compensation are less effective than they are now. Thus, if we use today's differences, found in statistical studies, between latitudes, both in DUV and skin-cancer rates, to forecast the eventual changes after 1, 2, or 3 decades in skin-cancer rates on the basis of an increase in DUV at one latitude, we will somewhat, but probably not greatly, underestimate the effects of the increased DUV. We may, also, if present behavior enhances latitude differences rather than partially compensating for them, somewhat overestimate these effects.
Melanoma incidence and death rates are rising rapidly in advance of any man-made disturbance of the ozone layer that has been measured. The increase in the death rate in the United States is somewhat over 3 percent per year (Lee et al., 1979). The increase in the incidence rate will probably be higher, since prognosis is improving, and earlier diagnosis of tumors is being achieved. A reasonable estimate would be for an increase in the current rate of increase from 3 percent per year to 4 percent or 4.5 percent per year for perhaps five years, followed by a reversion to the current rate of 3 percent annual increase. Recent epidemiologic studies of the National Cancer Institute have concluded that the Biological Amplification Factor (fractional change in skin cancer incidence rates per fractional change in UV-B radiation) is 2x, rather than 1x as previously presented.
Since the Physical Amplification Factor (fractional change in UV-B radiation per fractional change in ozone concentration) is -2, the overall amplification factor (fractional change in skin-cancer incidence-rate per fractional change in ozone concentration) is -4, rather than -2 as previously presented.
An increase in melanoma deaths is likely, but not certain, to occur as a consequence of a continuing increase in the rate at which DUV is received at the ground. Such a melanoma increase, if it occurred, would be delayed, well beyond the onset of a DUV increase, while the accumulated dose builds up on individuals. A 16 percent ultimate reduction in ozone, with a consequent 44 percent ultimate increase in DUV accumulation rate, might be expected to cause a 64 percent increase in melanoma deaths, if most melanoma deaths are solar-UV radiation related.
Estimates of percentage increases in skin cancer for various amounts of ozone reduction were given in the report of the Climatic Impact Committee (1975) (Table 3.4). Table 3.5 is a summary of more recent results compiled by Scott and Straf (1977).
CURRENT PROGRAMS OF THE NCI TO RELATE UV-B TO THE INCIDENCE OF MELANOMA AND NONMELANOMA SKIN CANCER
New information is being accumulated by the Biometry Branch of the National Cancer Institute to reduce the degree of uncertainty in the estimates of the dose-response relationships with UV-B. There are new surveys to determine the incidence of melanoma and nonmelanoma, to estimate the effects of UV-B, other environmental factors, and host (genetic) susceptibility from a geographic and demographic cross section of the country. In addition to these environmental epidemologic surveys, the NCI in collaboration with other agencies (e.g., EPA, NOAA, and NASA) is monitoring UV-B measurements being collected at locations where population-based cancer registers are in effect. The preliminary results from NCI's earlier studies were consistent with the hypothesis that increased amount of UV-B may lead to increased incidence of skin cancer. Some of the preliminary results from the current (1978-1979) surveys are as follows:
These studies are still in progress, and two more locations (San Diego and New Hampshire/Vermont) have now been added to the original sites (Seattle, San Francisco, Detroit, Atlanta, New Mexico, New Orleans, Utah, Minneapolis, St. Paul, and Bethesda).
The NCI is collecting skin melanoma incidence from the 10 locations in the United States. Earlier data analyses indicate that skin melanomas, which predominate in Caucasians, are related to UV-B exposure. A preliminary finding is that the incidence is increasing and that the rate of increase may be greater than expected from earlier surveys. The dose-response relationship between skin melanoma and UV-B appears to be more complicated than that observed for nonmelanomas of the skin.
New research studies are planned or in progress: (1) To ascertain the amount of UV-B reaching the skin by use of newly devised personal dosimeters; field studies are just beginning in contracts funded by EPA. (2) Skin Melanoma Case/Control Studies in new epidemiological data bases will provide relevant causative information, e.g., hormone therapy, oral contraceptive usage, exposure to short-term high-intensity solar radiation, occupational-industrial-chemical exposures, and residence mobility. These surveys are planned in two southern and two northern locations.
*Status report on research programs at the National Cancer Institute relevant to the health effects of UV-B radiation.
*The standardized mortality ratio simply compares the observed numbers of deaths with those expected for the age distribution of the occupation under study if the death rates from the particular disease in the whole population applied.
Auerbach, H. 1961. Geographic variation in incidence of skin cancer in the United States, Public Health Rep. 76:345.
Braun Falco, O. 1974 Maligne Melanome der Haut aus dermatologischer sicht. Chirurg. 45:345-356
Brodkin, R. H., A. W. Kopf, and R. Andrade. 1969. Basal-cell epithelioma and elastosis: a comparison of distribution. The Biologic Effects of Ultraviolet Radiation: With Emphasis on the Skin. F. Urbach, ed. Pergamon, New York, pp. 581-618
Camian, R., A. J. Tuyns, H. Sarrat, C. Quenum, and I. Faye. 1972. Cutaneous cancer in Dakar, J. Nat. Cancer Inst. 48(1):33-49.
Climatic Impact Committee. 1975. Environmental Impact of Stratospheric Flight: Biological and Climatic Effects of Aircraft Emissions on the Stratosphere, National Academy of Sciences, Washington, D.C.
Cutler, S. J., and J. L. Young, Jr. 1975. Third National Cancer Survey: Incidence Data. Nat. Cancer Inst. Monogr. No. 41.
Cutler, J., M. H. Myers, and S. B. Green. 1975. Trends in survival rates of patients with cancer, New Engl. J. Med. 293:122-124.
Elwood, J. M., and J. A. H. Lee. 1975. Recent data on epidemiology of malignant melanoma, Seminars Oncol. 2(2):149-154.
Fleming, I. D., J. R. Barnawell, P. E. Burlison, and J. S. Rankin. 1975. Skin cancer in black patients, Cancer 35(3):600-605.
Green, A. E. S., G. B. Findley, Jr., K. F. Dlenk, W. M. Wilson, and T Mo. 1976. The ultraviolet dose dependence of non-melanoma skin cancer incidence. Unpublished data.
Lee, J. A. H., and D. Strickland. 1976. University of Washington. Personally communicated data. Occupational Mortality Tables. 1971. H.M.S.O., London.
Macdonald, E. J. 1974. A statement on skin cancer incidence and sun exposure. Manuscript.
Magnus, K. 1977. Incidence of malignant melanoma of the skin in the five Nordic countries; Symposium on Solar Radiation, Int. J. Cancer 20:477-485.
Mason, T. J., and F. W. McKay. 1974. U. S. Cancer Mortality by County: 1950-1969. DHEW Pub. No. (NIH) 74-615. U.S. Govt. Printing Office, Washington, D.C.
McDowell, A. 1974. Nat. Center for Health Statistics, personal communication.
National Cancer Institute. 1974. The Third National Cancer Survey Advanced Three Year Report: 1969-1971 Incidence. Preprint.
Okono, A. N. 1975. Albinism in Nigeria. Brit. J. Dermatol. 92:485-492.
Registrar General's Statistical Review of England and Wales for the Three Yeras 1968-1970: Supplement on Cancer. 1975. Table E, p. 29.
Scott, E. L., and M. L. Straf. 1977. Ultraviolet radiation as a cause of cancer, in Origins of Human Cancer, Book A, Incidence of Cancer in Humans, H. H. Hiatt, J. D. Watson, and J. A. Winsten, eds., Volume 4 of The Cold Spring Harbor Conferences on Cell Proliferation, The Cold Spring Harbor Laboratory, New York, pp. 529-546.
Scotto, J., J. Franmeni, and A. H. Lee. 1976b. Melanomas of the eye and other noncutaneous sites: epidemiologic aspects, J. Nat. Cancer Inst. 56(3):489-491.
Scotto, J., A. W. Kopf, and F. Urbach. 1974. Nonmelanoma skin cancer among Causacians in four areas of the United States, Cancer 34:1333-1338.
Scotto, J., T. R. Fears, and G. B. Gori. 1976a. Measurements of ultraviolet radiation in the United States and comparisons with skin cancer data, DHEW Publ. No. (NIH) 76-1029.
Silverberg, E. 1979. Cancer Statistics, 1979, CA-A Cancer J. for Clinicians 29:6-21.
Sober, A. J., M. S. Blois, W. H. Clark, T. B. Fitzpatrick, A. W. Kopf, and M. C. Mihn. In Press. Primary Malignant Melanoma of the Skin--1130 Cases from the Melanoma Clinical Cooperative Group. In Proceedings XV International Congress of Dermatology, Mexico, October 16-22, 1977. Excerpta Medica.
Third National Cancer Survey 1969-1971: Incidence Data. 1975. National Cancer Institute Monograph 41.
Urbach, F., D. B. Rose, and M. Bonnem. 1972. Genetic and environmental interactions in skin carcinogenesis, Symp. on Fundamental Cancer Res., the 24th, M. D. Anderson Hosp. and Tumor Inst., Houston, 1971. Environment and Cancer: A Collection of Papers. Williams and Wilkins, Baltimore, Md., pp. 355-371.
Chlorofluoromethanes released into the atmosphere remain in the lower region, the troposphere, for many years and absorb radiation from the sun and also reradiated energy from the earth. The effect is a slight warming of the troposphere and a resultant warming of the earth's surface. This is known as the direct or greenhouse effect. Eventually the CFMs rise to the stratosphere, where they participate in the depletion of ozone. This process results in a cooling of the stratosphere, but the effect on surface temperature is small and its sign is in doubt.
This chapter describes the mechanisms that produce the climatic changes noted above and also estimates their magnitudes.
CLIMATE AND CLIMATE CHANGE
Climate is defined for purposes of this report by the statistical properties of weather over a portion of the earth or the whole earth. Climate may refer to any atmospheric level but usually refers to conditions near the surface of the earth. Although climate properly extends to measures of variability (such as ranges and standard deviations), we will concentrate here on averages (arithmetic means).
If we wish to study climatic change, we may look at differences between climatic averages based on two or more successive time intervals of observations. A portion of such differences will be due to the statistical uncertainty of averages taken over limited periods. Madden (NCAR, personal communication) has recently shown that a change between successive nonoverlapping 20-year averages of surface temperature at 60deg. N must exceed 0.4deg.C to be statistically significant (that is, to have a probability of less than 5 percent of arising from pure chance). Thus, climate differences must be larger than a certain minimum to be judged "real." To find the limiting differences is a purely statistical problem and has been attacked in different ways.
But after "real" climatic changes have been detected, the question must be answered whether they are human-induced or might have occurred through other causes. It is useful to study historical records of climate to try to understand the characteristics of past climate changes--when people's activities almost certainly had negligible effect. Only if observed changes significantly exceed normally expected changes can we be reasonably certain that human-induced changes have occurred.
Climatic changes have occurred principally over three broad ranges of time scales: hundreds to tens of millions of years, hundred to tens of thousands of years, and hundreds to tens of years. The longest of these scales is probably associated with continental drift and with associated changes in the distribution of land and sea. A noticeable part of the changes at the middle scale is associated in part with predictable changes in the earth-sun geometry. But the reasons for climatic variations on the century scale are unknown and certainly unpredictable.
For example, global average temperature is believed to have increased at the rate of about a degree per century from 1880 to about 1940. It has since decreased at about the same rate. Local changes, especially at high latitudes, have been much larger.
Fluctuations on even smaller scales have been of considerable interest. But here, the confusion between "weather" and climate is particularly serious. In order to identify human-induced climate changes in the next century or so, they must be significantly larger than changes expected naturally. But anticipated man-made changes, at least in the next half century or so, are of the same order of magnitude as "natural" changes. Therefore it would take considerable time to be sure about man-made climate changes by observation. The alternative is to predict such changes by physical models.
Physical models of the state of the atmosphere rely on numerical solutions of the basic meteorological equations. Most models refer to the atmosphere with its actual composition. But models have also been produced of atmospheres with altered composition. The most complex and realistic models are called GCMs (originally for general circulation models, but they alter also for global climate models). These models are three dimensional. The models are based on solutions of the seven basic equations of meteorology: the gas law, the first law of thermodynamics, the equations of continuity for air and moisture, and the three components of acceleration of Newton's second law. They are solved in small time steps (about 10 minutes) for the seven variables, pressure, temperature, density, moisture, and the three velocity components, subject to suitable boundary conditions. Of course, the earth's surface is a most complex boundary and has to be greatly simplified because of the limited resolution of the models. Typically, the horizontal distance between grid points is of the order 500 km, and, in the vertical, there may be 2 to 50 levels.
With proper heat sources and sinks, the models have reproduced many of the features of the real atmosphere, such as the general temperature distribution, the circulation, and statistical properties of weather. But because of the noise produced by the weather, the models, just as the real atmosphere, exhibited changes of averages from year to year. But as long as solar radiation, the composition, and the boundary conditions were held constant, these variations in the model were truly statistical.
In more recent GCM's, it was not necessary to begin integrations from an isothermal atmosphere. Integration time can be saved by beginning with a more realistic approximation of the real atmosphere. Even so, the computer time for simulation of much more than a year at 10-minute time steps for each of many grid points is formidable.
GCM's are by no means perfect. Clouds are simulated imperfectly, so is the interaction between air and sea. Scales smoothed out in the model affect the large-scale motion in a manner not modeled perfectly.
Climatic models have also been devised in two or one space dimensions. Such models save computer time and also permit better insight into the physical processes. But such models are limited by the difficulty of simulating horizontal transport. It is not at all clear whether models in less than three space dimensions are even qualitatively reliable. However, for certain changes in imposed conditions such as a doubling of CO2 or a change in solar constant, the 1-D and 2-D models agree, within 25 percent, with the GCM results with regard to computed change in surface temperature.
GCM's can be reconstructed with altered boundary conditions or atmospheric composition. If ozone changes are to be considered, an eighth equation for continuity of ozone has to be added. If the CO2 content is changed, infrared radiation fluxes are changed. Generally, the altered radiation budgets can be treated well, but, because of the imperfect description of feedbacks involving ocean or clouds, the response of the atmosphere is less certain. Further, any changes in the GCM model climate from its original state to its altered state must exceed significantly the statistical "noise" of the model climate. But, when considering man-made influences, the models have this advantage over the real atmosphere: changes due to changed composition or other features have to exceed only the statistical noise--not natural climate changes due to mostly unknown causes.
THE CO2 PROBLEM
At present, the most serious man-made impact on climate in the next few centuries is believed to be produced by the increasing concentration of CO2 in the atmosphere, as a result of burning fossil fuels. Global CO2 content has been measured since at least 1948 when a monitoring program was first established. Reliable observations are available only since 1958. There has been an increase of about 50 percent of the amount of CO2 released into the atmosphere by combustion of fossil fuels. Presumably, most of the other 50 percent has been taken up by the ocean. The mechanism for this uptake has been clarified quite recently. Whether changes in the biosphere contributed to the increase in CO2 is not clear. Although this report deals with potential effects of ozone on changes of climate, we will use potential CO2 effects, which are comparatively well understood, as a yardstick.
Figure 4.1 shows two alternative scenarios for the increase of CO2 in the next century. One scenario continues with the present increase of burning of fossil fuels, about 4.3 percent per year, multiplied by the fraction of carbon expected to remain in the atmosphere. The other assumes some controls, limiting the growth to 2 percent till 2025, followed by a symmetrical decrease thereafter. It is seen that, by the year 2050, the CO2 content may easily double, perhaps even triple.
Carbon dioxide is nearly transparent to most solar radiation in the visible region of the spectrum, where much of the sun's radiation is concentrated, but it has several strong absorption bands in the infrared, centered near 15um. Increasing the CO2 content in the atmosphere will increase the atmospheric IR opacity with a consequent enhancement in the atmospheric greenhouse effect, which leads to a warming of the surface and troposphere. At the same time, the stratosphere will cool because of the increase in CO2 IR emission.
Many quantitative models of the probable effect of CO2 change on climate have been made; Figure 4.1 shows the range of increases of global mean surface temperature. In particular, the best estimate for global average surface temperature change for doubling CO2 is 3deg.C with considerable uncertainty due to the difficulties of atmospheric modeling. In polar regions, however, the temperature increase is likely to be twice as large, in part because of the amplification of the CO2 warming by ice-albedo feedback. When the atmosphere warms as a result of any cause, ice melts; this reduces the ground's reflective power (albedo); more sunlight is absorbed, and the temperature is increased further.
Since polar regions warm more rapidly than tropical regions, horizontal temperature gradients will decrease, leading to reduced winds. At the tropical tropopause, the temperature should increase by 2-3deg.C, leading to increased water-vapor concentration in the stratosphere. In the middle stratosphere (30 km) the average cooling is expected to be 5deg.C (see Manabe and Wetherald, 1975). In addition to temperature changes, increase of CO2 should also change global regimes of other meteorological variables, e.g., wind and precipitation.
The statistical characteristics of natural climatic variations over decades are not well known, and it appears that the temperature changes due to increased CO2 will not be detectable as such before the year 2000.
DIRECT EFFECTS OF CFMs ON CLIMATE
The climatic effects of an increase in the atmospheric concentration of CFMs are twofold: (1) An increase of CFMs can perturb the climate directly by enhancing the atmospheric opacity in the 8-12um wavelength region and thereby contribute to the atmospheric greenhouse effect. (2) An increase of CFMs can also perturb the climate indirectly by altering the ozone amount or distribution, which will then affect the radiation balance.
Subsequent to the original publication by Ramanathan (1975) about CFM greenhouse effects, three publications, namely, those of Wang et al. (1976), Reck and Fry (1978), Boyer (1979), have presented one-dimensional model estimates for the global warming due to an increase in CFMs. All but one of these estimates, namely, that of Wang et al., agree within 10 percent of Ramanathan's estimates. Wang et al.'s estimate for the surface warming is smaller than Ramanathan's values by about 40 percent. Roughly half of the difference appears to be due to differences in treatment of radiative effects of CFMs; more detailed explanations are available from Ramanathan (1978). Boyer (1979) has estimated the CFM warming with a Budyko-Sellers type energy balance climate model. This model computes surface temperature change as a function of latitude. Boyer obtained warming rates about twice as large as those estimated previously by 1-D radiative-convective models. For a uniform mixing ratio of 0.7 ppb of F-11 and 1.9 ppb of F-12, Boyer estimates a global warming of 0.9deg.C compared with the 1-D model estimate of 0.42deg.C reported in the CISC (1976) report. He ascribed this change to the ice-albedo feedback mechanism. The approximate doubling of the global warming, i.e., from 0.42 to 0.9deg.C due to ice-albedo feedback, seems to be a significant overestimation in view of recent studies by Lian and Cess (1977) as well as Manabe and Wetherald's GCM experiments, which suggest that ice-albedo feedback amplifies global warming by a factor of about 1.25. Thus a figure of about 0.5 to 0.55deg.C for 0.7 ppb of F-11 and 1.9 ppb of F-12 seems indicated.
Temperature Changes above the Surface
A recent study by Dickinson et al. (1978) provides the only 3-D model estimates for the direct effects of CFMs. These investigators used the GCM of the National Center for Atmospheric Research. They estimated the response of the GCM climate due to the direct radiative effects of 10 ppb of CFMs and then scaled their results down to CFM concentrations inferred for indefinite continuation of 1975 CFM emission rates. However, Dickinson et al.'s GCM results cannot be used to infer CFM-induced surface temperature changes because the model prescribes changes in ocean surface temperatures (for the simulations with CFMs) in addition to prescribing ocean surface temperatures (for the simulations without CFMs).
The GCM results revealed an important aspect of the problem that was not brought out by the earlier 1-D model studies. They indicated that maximum warming due to CFM IR radiative heating occurs near the equatorial tropopause region. This maximum in warming is due to a corresponding maximum in the vertical distribution of CFM radiative heating. The tropical tropopause warming is estimated to be about 2.5deg.C, which exceeds the surface warming by more than a factor of 2. This effect is potentially important because the cold equatorial tropopause temperature is believed to be the most important factor that limits the water vapor that can enter the stratosphere from the troposphere. If this hypothesis is correct, a 2.5deg.C warming of the tropical tropopause may increase stratospheric water vapor concentration by as much as 60 percent. A 60 percent uniform increase in stratospheric H2O can have appreciable effects on ozone chemistry. In addition, the increased stratospheric H2O through its IR greenhouse effect can amplify the surface warming caused by CFM IR radiative effects.
CLIMATE CHANGES DUE TO OZONE CHANGES
Alterations in atmospheric ozone concentrations can influence the climate in two ways: (1) Ozone contributes to the atmospheric greenhouse effect in the 8-12um region and in addition absorbs solar radiation. Hence ozone perturbations can perturb the IR- and solar-radiation fluxes absorbed by the climate system. (2) Changes in stratospheric temperatures caused by ozone perturbations may be transmitted to the troposphere by radiation and dynamical coupling mechanisms.
Tropopause and Stratospheric Temperature Changes
Mahlman et al. (1978) describe a GCM experiment that examines the climatic effects of ozone reduction. Mahlman et al.'s GCM has 40 vertical levels extending from the ground to about 80 km. Because of this fine vertical resolution, the model is particularly suited to studying ozone effects on stratosphere-mesosphere climate. The model, however, is not designed for examining surface temperature changes, since, like the NCAR GCM used by Dickinson et al. for examining CFM effects, the model prescribes ocean surface temperatures.
Mahlman et al. estimate the response of the GCM to a 50 percent uniform reduction (from the ground to 80 km) in ozone. The model predicts significant temperature changes in the stratosphere and mesosphere with negligible changes in the lower troposphere. The stratopause (i.e., the 50-60 km region) cools by 23deg.C and the tropopause cools by about 12deg.C. Mahlman et al.'s calculations cannot be used directly to assess the temperature effects of ozone reductions resulting from CFM injection since the ozone reduction profile due to CFM injection (as predicted by 1-D Models) has considerable vertical structure as opposed to the uniform reduction assumed in Mahlman et al.'s experiments. For the CFM injection, maximum O3 reduction occurs above 30 km, and O3 reduction at the tropopause levels is a factor of 2 to 3 smaller than that above 30 km. Such a vertical distribution of O3 reduction would minimize the tropopause cooling, since large O3 reduction in the upper stratosphere would allow more solar radiation to be absorbed at the tropopause level.
Radiative equilibrium calculations by Ramanathan and Dickinson (1979) indicate that temperature decrease at the tropical tropopause due to ozone reduction is about 1-2deg.C. Note that this is counteracted by direct CFM warming! The net effect is of uncertain sign but is more likely to be warming than cooling.
As to effects of ozone changes on surface temperatures, Ramanathan and Dickinson (1979) have modified the conclusions of earlier CISC reports somewhat. These authors examined how ozone changes alter the radiative energy budget of the earth-troposphere system. They computed the perturbations to tropospheric radiative heating as a function of latitude and season, but their model calculations ignore the feedback effects of additional radiative heating and consequent changes in dynamical processes in the stratosphere.
The most significant aspect of their calculations is that it illustrates the limitations of earlier 1-D model calculations. The earlier 1-D model studies showed that reductions of ozone have two competing effects on the surface-troposphere radiative heating:
The two effects are of opposite sign and the same order of magnitude. Ramanathan and Dickinson's detailed analysis confirmed this conclusion but showed that with changed ozone amounts additional solar heating is largely deposited near the surface, whereas additional IR cooling is largely deposited near the upper troposphere. If, as might be the case in higher latitudes, the surface and troposphere are not coupled strongly by vertical mixing processes, such a heating distribution would simultaneously cause warming of the surface and lower-troposphere (-0.3 km) and upper-troposphere cooling. Furthermore, the magnitude of the surface warming in higher latitudes would also depend on the amount of horizontal coupling between low and high latitudes involving meridional transport of energy in the atmosphere and in the oceans. Hence the ozone-climate problem is considerably more complicated than envisioned earlier in 1-D models. The 1-D models assume an extremely efficient coupling between the surface and atmosphere both in the vertical and in the meridional directions, a poor assumption in midlatitude to high-latitude regions.
Ramanathan and Dickinson adopted the ozone reduction profile as computed by 1-D photochemical models for continued injection of CFMs at 1975 emission rates indefinitely into the future (the total column reduction is 19 percent) and obtained a hemispherical mean surface-troposphere radiative flux decrease of about 0.1 W m[-2] (watts per square meter). This net decrease results from a flux increase to the surface of about 0.4 W m[-2] (sum of solar flux increase and a relatively smaller decrease in downward emitted IR flux caused by decreased tropospheric O3) and a decrease of IR flux to the troposphere (primarily to the upper troposphere) of about 0.5 W m[-2]. As mentioned earlier, whether the net decrease, of 0.1 W m[-2], would cool or warm the surface depends strongly on the efficiency of surface-atmosphere coupling processes. Furthermore, since the next flux change is determined by a small difference between two relatively large competing effects, possible errors of about +/-50 percent in each of the two flux perturbations mentioned above can change even the sign of the net effect from a maximum decrease in the net flux, to the surface-troposphere system of about -0.5 W m[-2] to a maximum increase of about 0.3 W m[-2]. Hence, the change in net radiative flux to the surface-troposphere system, due to the 19 percent steady-state O3 reduction caused by CFM injections at 1975 rates, is estimated as
The sources of errors considered, in arriving at the +/-0.4 W m[-2] uncertainty limits, are (a) errors in model calculations arising from neglect of dynamical feedback processes in the stratosphere and from radiation model errors; (b) uncertainty in the net change caused by either a lack of coupling or an extremely efficient vertical coupling between surface-temperature system; and (c) errors in vertical distribution of O3 computed by 1-D pbotochemical models. For example, if the 19 percent O3 reduction were distributed uniformly from ground to stratosphere because of effects of circulation not allowed for explicitly in the 1-D photochemical models, the change in net radiative flux would be -0.25 W m[-2] instead of the -0.1 W m[-2] quoted above.
Troposphere-Stratosphere Dynamical Coupling
The troposphere and stratosphere are dynamically coupled by propagating planetary-scale waves. The planetary scale waves are longitudinal pressure perturbations with associated wind and temperature perturbations. These perturbations, generated within the troposphere, exist in a wide range of spatial and temporal scales. The energy of waves on planetary scales (i.e., those with wavelengths of the order of 5000 to 10,000 km) propagates upward into the stratosphere. However, the vertical and latitudinal temperature gradients within the stratosphere modulate this upward energy propagation. These temperature gradients can be altered significantly by ozone perturbations, as shown by the studies of Mahlman et al. (1978), among several others.
Bates (1977), using a simplified dynamical model, concluded that changes in stratospheric temperature gradients can (by altering the upward propagation of wave energy) produce substantial changes in poleward transport of heat within the troposphere.
However, more sophisticated dynamical models such as that of Mahlman et al. (1978) have not exhibited the strong sensitivity of tropospheric circulation to stratospheric changes as inferred from Bates' simple model. Hence we do not expect any important changes in surface climate as a result of the large changes likely in the stratosphere.
COMPARISON OF NET TEMPERATURE CHANGES DUE TO CFMs and CO2
We present below our current best estimates of various climatic effects of CFM increase. Furthermore, in order to put the CFM effects in better perspective, we compare the CFM effects with the climatic effects of CO2 increases that are expected in the next 75 years.
The future concentrations of CFMs that are used in our estimates are adopted from Chang's 1-D photochemical model calculations, as reported by the CISC Panel on Chemistry and Transport. Chang considers several scenarios for future CFM emissions. For our purposes here, we assume that the 1976 CFM emission rates are continued indefinitely into the future.
With respect to future increases in CO2 concentrations, we adopt the projections as reported in the Department of Energy report by Baes et al. (1976). The scenarios used by these authors were described above (Figure 4.1).
We have adopted future CO2 mixing ratios that are a linear average of CO2 concentrations predicted by the two scenarios. Table 4.1 shows the adopted values for the time history for the ground-level mixing ratios of CFMs and CO2 and for the change in O3 due to CFMs increase. The O3 change is assumed to be linearly proportional to change in CFMs.
Table 4.2 compares expected temperature changes between 1975 and 2050 due to changes of the CFMs and CO2, based on the scenarios described in Table 4.1.
The uncertainty limits were determined in two steps. First error limits in expected changes of the radiation fluxes were estimated. These are typically only of the order of 10 percent of the changes of the fluxes themselves. An exception is the change in the net flux due to change in ozone, because the change is a difference of fluxes of opposite sign and about the same magnitude. In this case, the uncertainty is far larger than the net flux itself. The error is so large, in fact, that the net warming due to the combination of direct and indirect CFM action has an uncertainty equal to the expected mean.
Given the change in radiation fluxes, we calculate the changes in temperature by atmospheric models. The simplest is based on the assumption of radiative equilibrium. This assumption yielded the smallest temperature changes in Table 4.2. The highest values are based on more complex models, allowing for various kinds of feedback. It turns out that all feedbacks considered so far are positive. One such positive feedback, the ice-albedo feedback, we have discussed already.
Another positive feedback comes about as follows. If radiation produces warming, evaporation is increased. Water vapor is added to the atmosphere. Now, water vapor is a good absorber of IR radiation but transmits sunlight reasonably well. Thus, increased water vapor means an increased greenhouse effect, hence a warmer surface air temperature.
The largest temperature changes in Table 4.2 were obtained by assuming that positive feedbacks increase the radiation-equilibrium temperatures by a factor of about 3.5.
Table 4.2 shows that the warming due to the actions of the CFMs is expected to be an order of magnitude smaller than that expected from the increased CO2. If the CFM emission is continued at its 1976 level, we will have about 10 percent less time to deal with the CO2 problem than we will have without the CFMs.
Baes, C. F., H. E. Goeller, J. S. Olson, and R. M. Rotty. 1976. The global carbon dioxide problem. Rep. ORNL 5194, Oak Ridge Nat. Lab., Oak Ridge, Tenn.
Bates, J. R. 1977. Dynamics of stationary ultra-long waves in middle latitudes. Quart. J. R. Meteorol. Soc. 103:397-430.
Boyer, G. J. 1979. The "greenhouse" effect of CFMs in a simple energy balance climate model. JOC-GARP Climate Modeling Conference Proceedings, May 1978.
CISC. 1976. Committee on Impacts of Stratospheric Change. Halocarbons: Environmental Effects of Chlorofluoromethane Release. National Academy of Sciences, Washington, D.C.
Dickinson, R. E., S. C. Lin, and T. M. Donahue. 1978. Effect of chlorofluoromethane infrared radiation on zonal atmospheric temperatures, J. Atmos. Sci. 35: 2142-2152.
Lee, J. A. H., G. R. Petersen, R. G. Stevens, and K. Vesanen, 1979. The influence of age, year of birth, and date on mortality from malignant melanoma in the population of England and Wales, Canada, and the white population of the United States. Am. J. Epidemiol. Dec.
Lian, M. S., and R. D. Cess. 1977. Energy-balance climate models: a reappraisal of ice-albedo feedback. J. Atmos. Sci. 34:1058-1062.
Mahlman, J. D., R. W. Sinclair, and M. D. Schwarzkopf. 1978. Simulated response of the atmospheric circulation to a large ozone reduction. WMO publ. 511.
Manabe, S., and R. T. Wetherald. 1975. The effects of doubling the CO2 concentrationon the climate of a general circulation model. J. Atmos. Sci. 32:3-15.
Ramanathan, V. 1975. Greenhouse effect due to chlorofluorocarbons: climatic implications, Science 190-50-52.
Ramanathan, V. 1978. Unpublished notes. (Can be obtained from author at NCAR, Boulder, Colo.)
Ramanathan, V., and R. E. Dickinson. 1979. The role of stratospheric O3 in the zonal and seasonal radiative energy balance of the earth-troposphere system. J. Atmos. Sci. 36:1084-1104.
Reck, R. A., and D. L. Fry. 1978. The direct effects of CFMs on the atmospheric and surface temperatures. Atmos. Environ. 12:2501-2503
Wang, W. C., Y. L. Yung, A. A. Lacis, T. Wo, and J. E. Hansen. 1976. Greenhouse effects due to man-made perturbations of trace gases. Science 194:685-690.