Physical Rationale

3.1 Stochastic Forcing

Stochastic forcing as originally suggested by Hasselmann (1976), is a mechanism that can generate low-frequency variations in the climate system. How does it work? The atmosphere is constantly in motion and, while we do not think of atmospheric motion as being decadal in nature, atmospheric motion can readily induce decadal and longer motions in the more slowly varying systems (such as the ocean) that are coupled to the atmosphere. Sarachik et al. (1996) have suggested the analogy of the tossing of a coin, which generates arbitrary long fluctuations depending on the number of tossings. In a coupled system, damping mechanisms prevent arbitrary long time

55 0 100 200 300 400


FIGURE 3 Simulated Northern Hemisphere temperature varatious for 300 years with the ECHAM4/OPYC3 coupled climate model. Annually averaged and 50-year low-pass filter.

scales from occurring. In addition to atmospheric motion, interan-nual forcing of irregularly occurring El Niño events can similarly generate suitable ultra-low-frequency fluctuations in the coupled system. I fully share the view of Wunsch (1992), who proposed that stochastic forcing could preferably be considered a null hypothesis for decadal to centennial variability unless proven otherwise.

Is it possible to reproduce this type of variability with a climate model? I will here show results from the Hamburg coupled ocean atmosphere GCM (Roeckner et al., 1999) using the present concentration of greenhouse gases. Figure 3 shows the results of a 300-year-long integration. It shows the variation of the Northern Hemisphere surface temperature as well as the 50-year low-frequency variability. As can be seen by directly comparing this result with Figure 2 the internal variability follows very closely the observational estimate by Mann et al. (1998). It therefore appears likely that the internal variability of the coupled climate system can give rise to the kind of variations that have occurred in the climate system from 1000 to 1900. However, the model cannot reproduce the accelerated warming trend of almost 1 K over the past 100 years. Another modeling experiment using a mixed layer model did not generate the large low-frequency variability as the fully coupled model. This suggests that the stronger El Niño events than normal (ENSO)-type phenomena which are realistically reproduced by the Hamburg GCM (Roeckner et al., 1999; Oberhuber et al., 1998) apparently are required to generate realistic low-frequency variability.

We may therefore conclude that stochastic forcing is the most likely explanation for the natural variability in the period prior to 1900. It is also concluded that stochastic forcing for this model cannot explain the large, sustained warming during the 20th century. Similar results have been obtained in other model studies (Manabe and Stouffer, 1997).

3.2 Solar Irradiation Changes

The forcing of the climate processes of the earth through radiation processes is, as far as we know, remarkably stable even if seen in a very long perspective. The variability of the solar irradiation cannot accurately be determined from earth-based observations since clouds, aerosols, ozone, and other radiatively active gases interfere with the solar beam in the atmosphere. Observations from satellites have only been available for some 20 years. Satellite observations reveal that solar irradiation varies on very short time scales as well as with the 11-year solar cycle. The magnitude of this decadal variation is 1-2 W m~2 compared to the solar constant of 1367 W m~2, a variation of 0.1%. The radiation which reaches the earth must be spread over the whole area of the earth (which is 4 times larger than the interception area) and the planetary albedo is 0.3 (mostly due to clouds), which means that the solar variations translate to a variability of about 0.2 W m~2. Numerical experiments undertaken by Cubasch et al. (1997) suggest that such a small forcing may not be detectable in the troposphere. The reason is presumably that the damping influence of the oceans cancels the positive and negative parts of the signal.

The question of longer periods of solar irradiance has been hotly debated in recent years. Such possible variations are inferred from historical records of variations in sunspots, the so-called Maunder Minimum in the late 1600 (Eddy, 1976), analogues with other sunlike stars, and paleo measurements of radioactive isotopes supposedly coupled to solar variations. Cubasch et al. (1997) forced a coupled climate model with data provided by Lean et al. (1995) as well as by Hoyt and Schatten (1993) for the period 1700 until the present. As would be expected, when the fluctuations are on a time scale of centuries or so, the model response broadly follows the forcing. The linear warming trend for the 100 years 1893-1992 is 0.19 and 0.17 K, respectively.

We may conclude that if the estimated variations in the solar forcing are correct, they can explain global temperature changes at a level of a few tenths of a degree, although the actual patterns differ between the two data sets and are different from the pattern provided by Mann et al. (1998). However, the main concern is that we currently have no observational evidence of any low-frequency variations in solar irradiation, since reliable data exist only for some 20 years. It is essential to stress that the available long-term datasets of solar forcing are based essentially on the hypothesis that the sun is an analogue to certain stars, which may show such characteristic variations in their irradiation. Therefore, our ability to say more about solar effects will crucially depend on obtaining reliable observations of the solar irradiation over longer periods.

If the low-frequency variations in the solar irradiation were correct, they could explain the climate variability in the period before 1900 or so, but because of the small amplitudes they cannot explain the rapid warming during this century. The solar forcing must therefore with high probability be excluded as the major cause of climate warming during the 20th century. Neither is the solar variability required to explain the variability of climate as documented over the last millennium, since this can be explained by internal variability of the climate system.

3.3 Volcanic Effects

Volcanic aerosols (mostly sulfate) have been suggested to have global effects on the climate when ejected in sufficient amounts into the stratosphere. If the aerosols do not enter the stratosphere, they will be rapidly removed by precipitation, and hence the effect on climate can probably be ignored. The major eruption from Mount Pinatubo on the Philippine Island Luzon on lune 15-16, 1991, provided an opportunity to quantify the effect fairly accurately. The eruption was one of the largest in the 20th century. It is estimated (Krueger et al., 1995) that 14-21 million tons of S02 were ejected into the stratosphere. The volcanic cloud moved eastward by some 20 m s-1, thus encircling the earth in 3 weeks, whereby S02 was converted into sulfate aerosols (Bluth et al., 1995).

In the first month most of the aerosol mass was located in a band between 20°S and 30°N, and then the cloud gradually spread to finally encircle the whole global stratosphere. Radiosonde observations as well as measurements from the microwave sounding unit (MSU) indicated a global stratospheric warming of about 2 K. The observations also suggested a cooling of the lower global troposphere and the surface of the earth by about 0.5 K (Dutton and Christy, 1992).

There have been several attempts to calculate the climate effect of Mount Pinatubo, for example, those by Hansen et al. (1992). Bengtsson et al. (1999) recently carried out an experiment with the MPI high-resolution coupled ocean-atmosphere model. In this experiment the aerosol clouds were introduced into the stratosphere month by month for over 2 years and the corresponding change in radiation was calculated by the model. As observed, a rapid warming occurred in the stratosphere and a corresponding cooling in the troposphere. In Figure 4 we compare the results with observed microwave radiation data from the polar orbiting satellite microwave sounding unit. The model results have been expressed in the same units as the MSU data. To ensure that the model-calculated results were representative, an ensemble with six different integrations was carried out. The figure shows a

Global lower stratospheric temperature anomalies

Global lower stratospheric temperature anomalies

Global lower tropospheric temperature anomalies

FIGURE 4 Observed MSU temperature, shown as dashed line, for channel 4 (top) for the period 1979-1997 and the equivalent for the simulations with Mt. Pinatubo and stratospheric ozone. The mean value obtained from the six realizations is denoted by the solid line, whereas the shaded area represents this value plus and minus one standard deviation of the individual simulations, respectively. The same for channel 2LT (bottom). Note the steady response in the stratosphere and the large variability in the lower troposphere, (from Bengtsson et al., 1999).

FIGURE 4 Observed MSU temperature, shown as dashed line, for channel 4 (top) for the period 1979-1997 and the equivalent for the simulations with Mt. Pinatubo and stratospheric ozone. The mean value obtained from the six realizations is denoted by the solid line, whereas the shaded area represents this value plus and minus one standard deviation of the individual simulations, respectively. The same for channel 2LT (bottom). Note the steady response in the stratosphere and the large variability in the lower troposphere, (from Bengtsson et al., 1999).

model integration from 1979, an integration where the observed stratospheric ozone data were also considered. The effect of the eruption of El Chichon in 1983 was not incorporated.

As can be seen, the predicted tropospheric cooling is rather close to the observed temperature reduction. It is also quite robust since very similar results were obtained from all integrations. The stratospheric warming is somewhat overpredicted. The effect of the eruption lasted 5 years and was apparently prolonged due to delayed effects of the oceans. In conclusion it seems that major volcanic eruptions will affect global climate, but the cooling effect disappears comparatively fast. Only series of major eruptions are therefore likely to cool the global temperature on decadal and longer time scales and thus probably could explain at least part of the variations in the climate as occurred over the Northern Hemisphere from 1000 until 1900 (Lindzen and Giannitsis, 1998).

However, the rapid warming during the last century can hardly be attributed to quiescent volcanic activity. Although a systematic decrease in volcanic activity from the late 19th and early 20th centuries may have contributed to the relatively fast warming in the 1930s and 1940s, it appears highly unlikely that reduced volcanic activity is the reason for the warming trend during the 20th century in general, with the 1990s as the possible warmest decade in this millennium. The fact that the most intense volcanic eruption this century occurred in 1991 makes it even more unlikely.

We may thus quite safely conclude that the systematically reduced volcanic activity cannot be the cause of the sustained warming trend in the 20th century. Through the process of elimination, this leaves us with anthropogenic effects as the most likely

3.4 Anthropogenic Effects

As has been well documented by IPCC (1990, 1994), the climate forcing since the beginning of industrialization including greenhouse gases (GHGs), aerosols, and land use has changed and continues to do so at an accelerating pace. Since the beginning of industrialization, the overall forcing from C02, CH4, N20, and CFCs has increased by some 50%, and more than half of this has occurred in the last 40 years. While the forcing from the well-mixed greenhouse gases is known with an accuracy of less than 10%, practically all the other forcing factors have considerable inaccuracies (Fig. 5). This is particularly the case for the indirect effect of aerosols, which is only known within error limits of 50-100%. The effects of vegetation changes and other anthropogenic surface alterations are equally poorly known and have not yet been properly investigated in realistic modeling experiments. It is clear that a important objective for the future will be to arrive at a more accurate determination of climate forcing than we presently have. To better understand the role of aerosols in climate forcing is particularly important.

IPCC has tried to estimate the future change in the atmospheric concentrations of the well-mixed greenhouse gases. Present projections used by modelers are essentially based on an extrapolation of the increase during the last decades (for C02 it is about

Greenhouse effect

CFCs n2o ch4

Tropospheric aerosols

direct indirect small

Strato- Tropospheric spheric ozone ozone

Uncertainties large large large very large

FIGURE 5 The direct greenhouse effect in W m 2 at the top of the tropopause from the beginning of industrialization until present. After IPCC.

1%/year). Despite considerable efforts to reduce the emissions, I would be surprised if it is possible to avoid a further increase by some additional 50% before the middle of the century. It is rather likely that it will increase even more. The improvement in present general living conditions in countries outside Europe, Japan, and the United States can hardly be accomplished without substantial increase in the use of fossil fuels.

Future changes in methane are difficult to estimate because we do not yet have complete knowledge of the sources and sinks of methane. The recent slowing down in the increase of methane is not well understood. Another problem concerns possible changes in the future carbon cycle. Will the future uptake of carbon in the terrestrial biosphere increase faster or slower than the emission of carbon dioxide or not? To better understand the carbon cycle in a changing climate is one of the big challenges of the future.

Was this article helpful?

0 0

Post a comment