Click for next page ( 86


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 85
l 10 -Global Temperature Trends KEVIN E. TRErIBERTH Notional -Center for Atmospheric Research This talk focuses mairdy on stratospheric temperature trends but covers some a~pC{B of Propose temp~at;ures ~ -weD. The I part of the talk 6tiSCUS$~S what is h~-~i-ng in the arc region. The principal Issue to consider ~ the onsmte-~ -of coons am other trace gas changes, and of the expected temperature Mange best ing hum such trace gas changes, with actual temperature Amps. In the Iowe:r stratosphere, espe~c~aIly Dublin the Hider night, !tem- perat;ure am ozone are simita~y at so :from that -$taMpomrt temperature an] sconce changes sh-4 -recur in-tandem. But o:zorre ;ch~ges also imply Ganges in Bow he~g rates and th~u~ch-anges in temperature. Are changes in ozone an;~peratwe`.~sm~tent that sense? With regard tc, greenhouse gayest, merry indicates that a warm- ~ng should occur at the surface, but since the oceans add the cryosphere are involved' cih~ges a" Alley to occur on time scales of a-~;ecade or ITIQ1~, whereas flee reuse of the ~tr~at~.ihere to greenhouse gases Ill c~cciur ran a mther shout tinge sold. For the stratosphere, `ozone decrease and g~eenho; - e .~" morale b`atth Icy GO:Oli ng; therefore the Ounce ~ d*~g an gun Change is greater than in the ~tr~.~e. Jerry mamas rheas mead Ridded out that c~omp-ara~y Platte, ;; Ding; ia-ave Occurred in the upper str;atos~ere, but e~ewhe;re temperature trencis care much harder to find. >85

OCR for page 85
l 86 KEVIN E. TRENBERTH One of the problems with temperature trends is the difficulty of detecting them. Temperature records are clearly functions of the measurement systems used. The issue of temperature measurements needs to be addressed in the future, especially with regard to any climate change program. The main device for taking temperature measurements in the atmosphere is the radiosonde, which was not, however, designed to detect long-term temperature trends. There have been changes in type and design (especially with regard to radiation shields), and significant day-night temperature differences that have varied with design have resulted. A recent international intercalibration found that instruments of different nations generally reproduce temperatures within about 0.2C and pressures within 2 mb. This amount of error is not serious in the troposphere but becomes significant in the stratosphere, where 2 mb is a sizeable fraction of the pressure being measured. Hence, radiosondes are not useful above about 20 mb; ~ fact, few of them ever rise higher than 20 mb. The other principal source of temperature data in the free atmo- sphere is satellite radiance data that exist in useful form (as TOVS data) from about 1979 to present. There are again problems with cal- ibration, changes to new satellites, and changes in the time of day of satellite overhead passage. These problems have led to temperature differences of 2 to 4C, due to changes of satellite in the stratosphere especially. Therefore, it has been necessary to adjust and merge segments of the satellite record in order to get a continuous record of temperature trends. Some of the analyses have in fact removed apparent (but spurious) trends by invoking this kind of step-function adjustment. So it is very difficult to get an absolute calibration and to use satellite radiance data to get reliable trends. Rocketsondes are too few and far between to be useful for estate fishing trends alone, but they may be useful for calibrating satellite data in the region above the useful radiosonde ascents. But rocket- sondes have also changed, in type and instrumentation, with time. The products that exist for determining temperature trends are station or instrument records and analyses based in some way on these records. Many station records suffer from missing data, a major problem. Additionally, the areal coverage is insufficient in many regions. Various kinds of analyses have been done to fill the record gaps and to merge different data types (e.g., Labitzke et al. (1986) have compiled temperature analyses for the Northern Hemisphere that start in 1965~.

OCR for page 85
GLOBAL TEMPERATURE l:RENDS 87 When one examines a temperature record for a linear trend, one needs to be concerned with rend effects" and with major temporary perturbations, such as those attributed to E} Chichon. The seasonal cycle, which has very large amplitude in the stratosphere, is impor- tant in that incomplete data during a period of rapid seasonal change can give misleading results, as explained below. The missing data problem arises because of balloon loss or data transmission problems. Because of differing data amounts at different levels, the thickness of the monthly mean geopotential heights may not equal the monthly mean of the daily thickness values. The effects on temperature trends that we need to consider in- clude the solar cycle with its variation in ultraviolet flux, leading to stratospheric cooling at the solar minimum; aerosols from volcanic eruptions, resulting in heating within the stratospheric aerosol layer; ozone change, with a decrease implying stratospheric cooling; carbon dioxide, with an increase implying stratospheric cooling; and natu- ral variability due to dynamical processes and internally generated variations such as the quasi-biennial oscillation, southern oscillation, and sudden stratospheric warmings, affecting both temperatures and the ozone distribution. The surface temperature record (Figure 10-1) as compiled by Hansen and Lebedeff (1988) shows that, in terms of global averages, 1987 was the second warmest year on record, being exceeded only by 1981. The 1980s thus far average distinctly warmer than any previous decade since the record began in the 1880s. It is noteworthy that global temperatures have been highest during a period of decrease in insolation as measured by satellite. The magnitude of the insolation decrease since 1979 is estimated to have offset almost exactly the expected increase since 1979 in temperature due to greenhouse gages. The trend of the 108-year record is not linear but has variations, including a relatively warm period in the 1930s, which has been attributed to a dearth of volcanic aerosols but could also be due to natural variability. There is a problem in interpreting this record because of such anthropogenic effects as development of urban heat islands (Kukla et al., 1986), deforestation, building of roads, and relocation of observing stations. Although an attempt has been made to estimate such effects, we cannot be sure that the record is indicating changes due to carbon dioxide buildup and other external factors. As ~ mentioned earlier, problems can arise in fitting linear trends to data. One can use least squares regression to fit a linear trend to

OCR for page 85
%8 0.4 o.~ _ ala --~.~2 _ I..] A. ,,, . .' At, . ~ lo.- ~ l , .+ , 'i, . -~.6` _4 . . r; ~ . 80 ! O. ~ E KEVIN E. TRENBERTH Annual Meon 5 Year :*~nning :Nlea2n . a. 1 1 . , 1 ! -+ - - - - - - - - - - ~i F I al [rfOr Estimates (95~0 Confidence] _ - i ~ :i 1 ! I . ! ~ I 1: : .1920,1~930 1~~O ;~-~50 .~1~960 '1970 ~1980 1-990 -''E FIG; 10-:1 Dp-atture of -mean g1~1 ~-temperat~res from their i951 to I-98Q period mean -~iiue for ~ndi:Vidua1 yews boots con~ect~1 With dashed Elmer and for 5-year running ~mea-n valu-es .~anlid ~unre3. ~rar~estim-~es for both in~ndll-al ~anii1 !tu~ining mean val;ues a-re s;}~o~tn i-= sil~.~d yeiars. ~lod -~t recc,rd ~s 1880 through ^~. `~pr~i~ed ~tro-m Ha-n~n ta~ ~efE, 1988. t~opy-r~-ght :~) 198-8 ;b.y the ;Pt-meric-an Oe'opb - ~1 C.ni-o-n-.~) o4ie cyche ~om pi 't0 thr-ce :pi) ~ -a sine wave ~;d get a lar-ge :pos~tive -trend w~kh a t~a~- on ~ 0.7-8, Wh~Gh ~ ~y ~s~:gn~fica~t. :If hadf a tyole is added ;~43 th-e tegmn~ng <3f-this ~record and a~her -half ;to $he en6, a ]=ge 8:oWnwat~ iindar -tPe~n:6 -~re~s ~hai h"as --a correiiatio~ ~f --~.~3!~. The tm~t ~of tllis is th~at-trui~ds ~re wry sensiti~re to ~-the be- -gt0tngand~nil-po~n-ts`-ofilietacotd. Co-ncli~sions~abo~;i OCR for page 85
GLOBAL TEMPERATURE TRENDS 89 value since the 1920s. Any trend starting with 19" is likely to be totally unrepresentative of longer-period trends. Karoly (1987) has looked at temperature trends for 19 Southern Hemisphere stations from 1964 to 1985. He found slight upward trends, ranging from 0.1 to 1.2C, at 700 mb at all of these stations and slight downward trends, ranging from-0.1 to -1.3, at 50 mb at all but one of them. The Mt. Agung volcanic eruption that occurred in 1963 may have affected the trends to some extent; hence the trends may reflect more than just the carbon dioxide increase. AISQ in the Southern Hemisphere, Salinger (1979) has composited New Zealand mean annual and seasonal surface temperatures from 1853 through 1975. A minimum in the rn~-1960s is even more pronounced in this record than in the global record of Hansen and Lebedeff (1988), and a marked upward trend since about 1964 is not representative of the overall long-term trend of this record. Labitzke et al. (1986) have published time-series of monthly mean Comb temperatures for the Northern Hemisphere from 1965 through 1983. The E! Chichon eruption of 1982, and perhaps the Mt. St. Helens eruption of 1980, contarn~nated the record in the last few years. Therefore Labitzke computed linear trends from 1966 through l9X0. These trends are slightly downward for this period at latitudes poleward of 20N, but the time series show much interannual variability, even though the quasi-biennial oscillation has been removed from these series. Only the trend at 70N latitude appears large enough to be meaningful. Using a relatively sparse network of stations, Angell (1986a) has looked at global temperature trends from 1960 to 1985. HE results show a slight warming in the Southern Hemisphere midtroposphere and a slight cooling above about 10 km, in general agreement with Karoly's results. These trends are somewhat stronger for the decade 1975 to 1985 alone. More recent results of Angell (unpublished) show layer mean temperature time-series for the world through 1987. The records for the surface, the 850~ to 300-mb layer, and the 300 to 100-mb layer start in 1957, and the 10~ to Comb record starts in 1969. The surface and tropospheric layer both show cooling un- til about 1964 to 1965 and a warming trend since. The 300 to 100 mb layer does not show a recognizable trend. At 100 to 50 mb, Angell's data show a pronounced downward trend after a warm peak in 1982 that was probably associated with the eruption of E! Chichon. Angell (1986b) did a latitudinal breakdown of the 10() to Comb data, which shows that a recent rapid decline of 7C in the South Polar

OCR for page 85
90 KEVIN E. TRENBERTH region accounts for much of the globally averaged decline, with the equatorial zone also making a significant contribution. The Ozone Trends Pane! (Watson et al., 1988; Trenberth, 1988) did a comparison of temperature time-series by Angell (1986b; 100 to 50 mb), Labitzke et al. (1986; 100 to 50 mb, for the Northern Hemisphere only), the National Meteorological Center (NMC) (70 to 50 mb, based on NMC analyses), and the microwave sounder unit (approximately 90 mb). There is good agreement for the Northern Hemisphere from 1979 through 1986, showing a warm peak in 1982 to 1983 that is probably due to E! Nino or E! Chichon, followed by a slight cooling trend. In the Southern Hemisphere, Angell's values indicate much colder temperatures than those for the other two series in 1985 to 1986. Angell's series also indicates the coldest temperatures in recent years in the 30N to 30S latitude region. A similar comparison for the 5- to 1-mb layer, based on the NMC analysis and two SSU channels, shows a clear downward trend in temperatures from 1979 to 1986 for all major regions of the world. For the antarctic region, time-series from 1979 to 1986 of 200-mb heights generated in analyses by the NMC and the European Centre for Medium-Range Weather Forecasting (ECMWF) analyses disagree by 50 to 400 m (Trenberth and Olson, 1988~. The obser- vational input to these analyses has an accuracy of about 20 m, for comparison. The ECMWF analysis has been getting progressively better with time. The NMC values were highly erroneous between 1983 and 1986 but improved in May 1986, when a correction to the analysis procedure was made. Clearly, the operational analyses in the antarctic region need further improvement. The record of radiosonde ascents at the Amundsen-Scott South Polar station has been examined and reveals that for the period 1961 through 1976, about 80 to 85 percent of ascents reached the 100-mb level during the winter (June to August) months, with about 40 percent surviving to 50 mb. For the period 1976 to 1986, on the other hand, only about 15 percent of these soundings reached 100 mb or higher, and only about 10 percent managed to reach 50 mb. The reason for this loss is the bursting of balloons in cold temperatures, so that a warm bias is likely to result from the remaining observations that are actually made. The situation is even a little worse at the McMurdo station. The worst month, in terms of the number of observations made, is August, but by October the amount of high- altitude radiosonde data obtained is much greater. This problem has a marked impact on the temperature records.

OCR for page 85
GLOBAL TEMPERATURE TRENDS 91 If we look at the thumb, thumb, 5() mb, and 30-mb annual cy- cles of temperature at Amundsen-Scott (Figure 10-2), we note that there is a very steep increase in temperature from August to Octo- ber. At 50 mb, the average temperature is -80C at the beginning of October and -55C at the end of the month. If there is a pre- ponderance of missing data at the beginning of the month, then the monthly mean temperature will have a warm bias. To date, analysis of the records of antarctic stations has not taken this problem into account. As a result, trends in temperatures deduced from antarctic station data are of questionable value. We have computed the departure from the mean annual cy- cle (Figure 10-2) of every individual day and then compiled these anomalies into monthly means for the antarctic stations. For 50 mb at McMurdo Sound, this compilation, when compared with straight averages of the available observations for each month, gives monthly means that differ by up to 12C for some Octobers. For recent years, our method gives mean October temperatures that are consistently up to 8C colder than those obtained by the straight averaging method. The adjusted October record since 1956 shows a slight downward trend that is not apparent in the unadjusted record. The above procedure was used for 3~, 5~, and 10(Ymb data at both Amundsen-Scott and McMurdo to correct for the annual cycle-rn~ssing data effect (see Figures 1~3 and 1~4~. For the month of September, there is little apparent trend from 1961 to 1987 at Amundsen-Scott or from 1956 to 1986 (with some years missing) at McMurdo. For October (Figure 10-3), a downward trend, more pronounced at Amundsen-Scott, is apparent in the last 8 to 10 years of these records, with the coldest years occurring near the end of the record. Even so, the recent downturn is modest and would probably not be considered especially noteworthy were it not for the known springtime ozone depletion in the antarctic region. For November (Figure 1~4), the recent downward trend ~ again appar- ent at Amundsen-Scott for 50 and 100 mb and at McMurdo for 100 mb only. It Is noteworthy, however, that at McMurdo the coldest November temperatures at all levels occurred in 1959 and 1961 (1960 data are maskings. For 1987, we have recently estimated the anoma- lies, shown by dashes in Figures 1~3 and 1~4, at the South Pole, and the lowest values on record occur in October and especially in November, consistent with the absence of heating due to the ozone hole.

OCR for page 85
92 KEVIN E. TRENBERTH -20 -22 -24 -26 -28 -30 -32 o -34 -36 -38 -40 -42 -44 -46 -48 -50 -20 -30 -35 -40 -45 - o -50 AL 55 -60 -65 -75 -80 -85 -90 RAW ANNUAL CYCLE WAVES 0-4 500 MB 92 183 274 365 RAW ANNUAL CYCl WAVES 0-4 \ 1 00 MB 92 183 274 365 DAY OF YEAR FIGURE 10-2 Mean annual cycle of temperatures at four pressure levels as observed at the Amundsen-Scott antarctic station. Heavy curve shows the daily raw means; thin curve is the best fit of combined temporal waves 0 to 4 to the daily means. Period of record is 1961 through 1986.

OCR for page 85
GLOBAL TEMPERATURE TRENDS -20 -28 -36 -44 -52 o -60 Q ~ -68 LD -76 -84 -92 -100 -20 -28 -36 -44 -52 o - -60 -68 -76 -84 -92 -100 RAW ANNUAL CYCLE WAVES 0-4 50 MB J f ,. '. _ I 92 183 274 365 l -RAW ANNUAL CYCLE 30 MB WAVES 0-4 per, i. ~ FIGURE 10-2 (continued). 93 92 183 274 365 DAY OF YEAR

OCR for page 85
94 KEVIN E. TRENBERTH MONTHLY MEAN T-ANOM 20 AMUNDSEN-SCOTT ..... , , . 15 10 o llJ -10 -15 -20 -25 /VAVAV \~1 LI V\\. HA HA . , v ~ . ~ v X ~4 30 MB; +150FFSET CJM = 6.612 - 50 MB GM = 4.574 100 MB: -15 OFFSET OM = 2.832 -30 . , . I 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 YEARS OCT MONTHLY MEAN T-ANOM ~MCMURDO ~ _ 15 10 - 5 o - O ,_ -5 -10 -15 -20 -25 \ /\ V 1 1? (JM = 8.646 50 MB (IM = 7.286 _ 100 MB; -15 OFFSET HIM = 4.784 30 MB; + 15 OFFSET \ ~ \ Al \ A MA ~ ~J: . ... . . . -30 - . , . , . , . , . 5 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 YEARS OCT - 30 - 50 100 -30 - 50 100 FIGURE 10-3 Monthly mean departure of daily temperature from the period normal for three pressure levels at Amundsen-Scott (top) and McMurdo (bot- tom) for October. Note that curves for the three different levels are offset from one another by 15C intervals. Data are missing for some years at McMurdo. The 1987 values for Amundsen-Scott are preliminary. See text for further explanation.

OCR for page 85
l GLOBAL TEMPERATURE TRENDS MONTHLY MEAN T-ANOM AMUNDSEN-SCOTT 30 30 MB; +15 OFFSET OM = 3797 25 - 50 MB cam = 5.066 ~ 00 MB; -15 OFFSET AM = 6 339 20 A 15 10 5 o - ~ O LL -5 -10 -15 -20 -25 1 - i~V' ~J~ it 1 1~i ~ 1','. 1 1 At\ -30- ., ., . , I 5 ~ 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 YEARS NOV MONTHLY MEAN T-ANOM 30 25 20 15 10 -10 MCMURDO - 30 MB; + 15 OFFSET aM = 4 915 50 M8 Cam = 4.851 100 MB;-15OFFSET (7M = 5994 ~ ~ \1: AN 1: -20 \ I -30 - ........ 5 6 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 YEARS NOV MA I,,/\ - 30 ~50 100 . . . . A so - 100 95 FIGURE 10-4 Monthly mean departure of daily temperature from the pe- riod normal for three pressure levels at Amundsen-Scott (top) and McMurdo (bottom) for November. Note that curves for the three different levels are offset from one another by 15C intervals. Data are missing for some years at McMurdo. The 1987 values for Amundsen-Scott are preliminary. See text for further explanation.

OCR for page 85
96 KEVIN E. TRENBERTH Finally, the Japanese antarctic station Syowa has measured both Womb temperatures and total column ozone from 1966 through 1985 (Chubachi, 1986~. For October and November, a distinct downward trend in both temperature and ozone is apparent after about 1978. December data show a slight downward trend in recent years. The year-to-year variations in temperature and ozone are approximately parallel to each other, supporting the argument that ozone is the key factor affecting temperatures at this level through solar heating. Syowa 50-mb November temperatures also show a downward trend from about 1978 on. Syowa is located close to the region where the largest trends are supposed to be occurring. (In answer to a question): The global surface temperature trends that have been computed are weighted toward the Northern Hemi- sphere, especially for the earlier years. Surface data coverage in the Southern Hemisphere before World War I! was only about half the present coverage, and there were no useful antarctic data prior to about 1954. REP1:RENCES Angell, J.K. 1986a. Annual and seasonal global temperature changes in the troposphere and low stratosphere, 1960-1985. Mon. Weal Rev. 114:1922- 1930. Angell, J.K. 1986b. The close relation between Antarctic total-ozone depletion and cooling of the Antarctic low stratosphere. Geophys. Res. Lett. 13:1240- 1243. Chubachi, S. 1986. On the cooling of stratospheric temperature at Syowa, Antarctica. Geophys. Res. Lett. 13:1221-1223. Hansen, J., and S. Lebedeff. 1987. Global trends of measured surface air temperature. J. Geophys. Res. 92:13345-13372. Hansen, J., and S. Lebedeff. 1988. Global surface temperatures: Update through 1987. Geophys. Res. Lett. 15:323-326. Karoly, D.J. 1987. Southern Hemisphere temperature trends: A possible greenhouse gas effect? Geophys. Res. Lett. 14:1139-1141. Kukla, G., J. Gavin, and T.R. Karl. 1986. Urban warming. J. Climate Appl. Meteorol. 25:1265-1270. Labitzke, K., G. Brasseur, B. Naujokat, and A. De Rudder. 1986. Long-term temperature trends in the stratosphere: Possible influence of anthropogenic gases. Geophys. Res. Lett. 13:52-55. Salinger, M.J. 1979. New Zealand climate: The temperature record, historical data and some agricultural implications. Climatic Change 2:109-126. Trenberth, K.E. 1988. Review of "Executive summary of the Ozone Trends Panel Report." Environment 30:25-26.

OCR for page 85
GLOBAL TEMPERATURE TRENDS Trenberth, K.E., and J.G. Olson. 1988. An evaluation and intercomparison of global analyses from the National Meteorological Center and the European Centre for Medium Range Weather Forecasts. Bull. Am. Meteorol. Soc. 69:1047-1057. Watson, R.T., M.J. Prather, and M.J. Kurylo. 1988. Present State of Knowl- edge of the Upper Atmosphere 1988: An Assessment Report. NASA Reference Publication No. 1208. National Aeronautics and Space Admin- i~tration, Washington, D.C.