Solar variability and climate

Transcription

1 Weather Vol. 55 Solar variability and climate November 2000 Joanna D. Haigh Imperial College of Science, Technology and Medicine, London The fundamental source of all energy in the climate system is the sun so it seems reasonable to assume that variations in solar output might be reflected in climate records. In this paper recent work on variations in total solar irradiance will be reviewed, together with its relationship to other measures of solar activity, and how past values may be reconstructed. Some studies concerning correlations between solar activity and atmospheric temperature will be discussed and results presented from a new multiple regression analysis of the relationship between 30 mbar geopotential heights and solar activity. The magnitude of the response cannot be explained purely in terms of direct changes in solar irradiance; other factors which might amplify the effect are discussed. Observations of solar variability Since ancient times it has been observed that the number of sunspots visible on the solar surface has varied. Records dating back to the early seventeenth century show a periodicity of 9-13 years in sunspot numbers; during the minima of this 1 1 -year activity cycle sunspots disappear while the number during solar maxima varies greatly. It has sometimes been speculated that changes in the weather might be associated with these variations (e.g. Herschel 180 l), but ground-based measurements of solar radiative energy (or total solar irradiance, TSI) showed no detectable variation. This led to use of the expression solar constant and an acceptance, at least by the meteorological establishment, that solar influence on climate on decadal-to-century timescales was negligible. The advent of satellite-borne radiometers has, however, enabled more accurate measurements of TSI to be carried out, and Fig. 1 shows the set of measurements of TSI made from satellites since From this it is immediately clear that the absolute value of TSI is not known to better than a few watts per square metre. However, closer inspection of 1374 MAX MIN MAX MIN NT E Fi YEAR Fig. 1 Measurements of total solar irradiance (TSO made between 1979 and 1999 by seven satellite instruments (courzesy Judith Lean) 399

2 the individual data records shows a consistent picture of variation of about 1.4Wmp2, or approximately 0.1 %, from minimum to maximum of the 1 1-year cycle. Although individual instrument records last for a number of years, each sensor suffers degradation during its lifetime so that construction of a composite series (or best estimate) of TSI from overlapping records is a complex task. Two recent studies demonstrate this: Willson (1 997) deduced that TSI was 0.5Wmp2 higher during the solar minimum of 1996 than during the solar minimum of 1986, while Frohlich and Lean (1 998) showed almost identical values in these two years. Other indicators of solar activity have been observed or measured at ground level and have records dating back over various periods. These include solar radio flux (measured at a wavelength of 10.7 cm) dating back to 1947, the aa geomagnetic index, which gives a measure of the magnitude of the solar magnetic field at the earth (measurements back to 1869), and solar diameter (measurements intermittently since 1683). Another indicator of solar activity is the flux of galactic cosmic rays (GCRS) reaching the earth, which is greater at times of lower solar activity; measurements of this flux have been achieved using groundbased ionisation chambers since Further indicators of longer-term solar variability have been derived from sunspot numbers, e.g. solar cycle length and solar cycle decay rate. Other terrestrially based indicators of solar activity recorded by cosmogenic isotopes in tree rings and ice cores also show longer-term variability. These various proxies vary markedly as indicators of solar activity. For example, over the past century sunspot number and solar radio flux at 10.7cm showed highest values at the solar maximum of 1958 while the aa index peaked during This is because whereas sunspot numbers return to essentially zero at each solar minimum the aa index shows 11- year cycles imposed on a longer-term modulation (Lean and Rind 1998). GCRs also show some differences from the sunspot record, e.g. GCR decreases tend to occur 1-2 years after the corresponding sunspot peaks and the magnitude of the decreases does not vary as the magnitude of the sunspot maxima. Clearly each 400 proxy is indicating a different aspect of complex changes in the sun and, if there are links between solar activity and terrestrial climate, the physical mechanism inducing that relationship should be reflected in some proxies better than others. At present the most plausible mechanism is that variations in total energy input are responsible so that, in order to understand the impact of solar variability on past climate, it is necessary to know past variations in TSI. Reconstruction of past TSI As direct measurements of TSI are available only over the past two decades it is necessary to use the other proxy measures of solar output to deduce (reconstruct) variations at earlier dates. In the simplest type of reconstruction a single indicator, such as sunspot number (Stevens and North 1996) or solar diameter (Nesme- Ribes et al. 1993), is calibrated against recent TSI measurements and extrapolated backwards using a linear relationship. A somewhat more physical approach recognises that solar radiative output is determined by a balance between increases due to the development of faculae (bright patches on the sun s surface) and decreases due to the presence of sunspots. Longer-term changes are also speculated to be occurring in the quiet sun against which these variable active regions are set. The sunspot darkening depends on the area of the solar disc covered by the sunspots, while the facular brightening has been related to a variety of indices. These include sunspot number (Lean et al. 1995), ultraviolet emission (the singly ionised calcium spectral line at nm, Lean et al. 1992), solar cycle length, solar cycle decay rate, solar rotation rate and various empirical combinations of all of these (Hoyt and Schatten 1993; Solanki and Fligge 1998). Lockwood and Stamper (1 999) use an entirely different approach to TSI reconstruction, based not on sunspot numbers but on the aa geomagnetic index. Figure 2 (a) shows group sunspot numbers from 1610 to 1996 (Hop and Schatten 1998), together with five TSI reconstructions. The sunspot numbers (scaled to correspond to Nimbus-7 TSI observations for ) show

3 I year Fig. 2(a) Reconstructions of wtal solar itradiance (Tsr) fiom: group sunspot numbers scaled w Nimbus-7 measurements (dash-dot curve), Lean et al. (1995) (solid curve), Hayt and Schatten (1993) sh$ed to match Nimbus-7 measurements (bold solid curve), Solanki and Fligge (1998) (dotted curves), Lockwood and Stamper (1999) (bold dash-dot curve). (b) Reconstructions of surface temperature: global average anomalies relative w average (Jones et al. 1999) (bom solid curve), Northern Hemisphere average (Mann et al. 1998) shtjied w match Jones et al. record in overlapping period (solid curve). little long-term trend. Lean et al. (1995) determine long-term variability from the sunspot cycle amplitude while Hoyt and Schatten (1993, shifted to match Nimbus-7 observations in 1986) use mainly its length. The two Solanki and Fligge (1998) curves are similar in derivation to the Lean et al. and Hoyt and Schatten methods but make an additional assumption concerning the long-term contribution of the 'quiet sun' based on observations of the behaviour of sun-like stars (Baliunas and Jastrow 1990) and the assumption that during the Maunder Minimum (an extended period during the late seventeenth century during which no sunspots were observed) the sun was in a non-cycling state. Lockwood and Stamper (1 999) predict somewhat larger variation over individual cycles but less on the longer term. Clearly, even disregarding the shifts due to absolute scaling, there are large differences between the TSI reconstructions such that the variation in solar radiative input to the earth is not well established. Climate response (decadal to century time-scales) Numerous papers have been published apparently presenting correlations between various measures of solar activity and different climate parameters; Hoyt and Schatten (1 997) present a good review of the more recent work. Many of the claimed results are, however, statistically questionable, sometimes because the data series used are not suficiently long or because other factors, such as the El Niiio Southern Oscillation or volcanic eruptions, which might have contributed to climate variability, have been ignored. One of the best established meteorological records is the global mean surface temperature series of Jones et al. (1999). This is shown, as an anomaly relative to the average, as the bold curve in Fig. 2 (b). The other curve shows the reconstruction of Northern Hemisphere surface temperatures of Mann et al. (1998), shifted to best match during the over- 40 1

4 lapping period. A visual comparison of the temperature records with the TSI reconstructions in Fig. 2 (a), e.g. increases in TSI coinciding with warmings during the first half of the eighteenth century and the early twentieth century, suggests a possible relationship but an objective analysis is necessary to confirm or deny this. The simplest approach to such an analysis is to consider the effects of direct solar radiative forcing. The TSI reconstructions shown in Fig.2 suggest an increase of W mp2 from the Maunder Minimum in solar activity of the late seventeenth century to the 1 1-year solar cycle minimum of This corresponds to a radiative forcing* of about W m- which, using a climate forcing parameter of degc Wp ' m2 (Intergovernmental Panel on Climate Change 1995), would suggest a solar-induced warming in global average surface temperature of approximately degc since that time. The temperature record shows a warming of about 0.7 degc over the same period, or about twice that indicated by the simple estimate above. More detailed statistical analyses, also assessing the contributions of other factors potentially causing global temperature change, and taking into account the natural variability of the system, have been carried out by several authors. For example, Mann et al. (1998) found significant detection of solar radiative forcing in the temperature record from the mid-seventeenth to early eighteenth centuries (the Little Ice Age period coinciding with the Maunder Minimum in sunspot activity) and *Geometric factors affect the conversion from change in TSI to radiative forcing: it is necessary to divide by a factor of 4, representing the ratio of the area of the earth's disc projected towards the sun to the total surface area of the earth, and to multiply by a factor of 0.7, to take account of the earth's albedo of 30%. Thus a variation of 1 W m in TSI represents a variation in global average instantaneous radiative forcing of W m '. Here radiative forcing has been defined as the change in net downward radiative flux at the top of the atmosphere; this is not the complete definition of Intergovernmental Panel on Climate Change (1995) which uses the value at the tropopause allowing for adjustment of stratospheric (but not tropospheric) temperatures. The effects of changes in stratospheric ozone on the adjusted radiative forcing are discussed below. 402 from the early nineteenth to the mid-twentieth centuries (during which both global temperatures and solar irradiance increased). A detectiodattribution study, in which patterns of change derived from general circulation model (GCM) responses to different forcings (greenhouse gases, sulphate aerosol, volcanic aerosol, TSI) were combined to give a best fit to observed patterns of temperature variation in the twentieth century, also suggests a detectable solar response (Tett et al. 1999). Both these studies conclude, however, that warming in the late twentieth century cannot be ascribed to solar variability alone but is more likely to be due to anthropogenic influences. Gillett et al. (2000) have also looked at variations in the vertical temperature structure using gridded radiosonde observations over the past 38 years and detected a significant solar contribution within this recent period. They suggest that taking the vertical structure of temperature changes into account might aid climate change detectiodattribution studies. Certainly the stratospheric response to increases in solar irradiance is likely to be different to its response to increasing greenhouse gases (Cubasch et al. 1997), and modelling studies (Haigh 1996, 1999; Larkin et al. 2000) also suggest changes in the strength and extent of the tropical Hadley circulations with concomitant changes in temperature. Time-dependent coupled ocean-atmosphere GCM simulations of surface temperature response to solar variability (e.g. Cubasch et al. 1997; Rmd et al. 1999; Tett et al. 1999) show global average changes consistent with the above energy balance estimates but with considerable variation across the globe. This stresses the potential pitfalls of drawing global conclusions from correlations between local climate parameters and solar variability. For example, the lower temperatures of the Little Ice Age were largely confined to Eurasia (Bradley and Jones 1993), suggesting that changes in atmospheric or ocean circulations might be implicated. Climate response (1 1-year cycle) Some of the most interesting studies purporting to show a signal of the 11-year cycle in

5 meteorological records have been carried out using mid-stratospheric 30 mbar geopotential height (Z30) data. Van Loon and Labitzke (1998), using data Erom both the Free University of Berlin (FUB) Analyses and the NCEPI NCAR* Reanalysis, show a relationship between Z30 and the 10.7cm solar index over nearly four solar cycles with the strongest correlations, of over 0.7, occurring in summer mid latitudes. These correlations are intriguing and it is instructive to calculate the implied magnitude of the response in 230 to a specified solar forcing. Figure 3 (back cover) shows the results of a multiple regression analysis of the FUB data ( ) in which indices representing greenhouse gas warming, volcanic aerosol and the Quasi-Biennial Oscillation (QBO) have been included along with solar activity and an autoregressive noise model to account for unexplained variance. The plots show the change in June Z30 deduced in response to: (a) an increase in solar activity of cm flux units (note that an increase of 1300 units took place between solar minimum in 1986 and maximum in 1990), (b) a typical QBO cycle, and (c) a volcanic eruption such as that of Mt. Pinatubo in This confirms the conclusions of Van Loon and Labitzke (1998) concerning the solar effect and shows a maximum solar signal of about 100 m over the subtropical Pacific Ocean. The volcanic signal compares in magnitude with the solar signal and also spreads over low to mid latitudes. A statistical analysis of the multiple regression results suggests that the solar and volcanic signals in mid to low latitudes (but not high latitudes) are significant at the 95% level. However, the fact that the eruptions of Mt. El Chichon in 1982 and Mt. Pinatubo in 1991 both occurred at about the same phase of two of the (nearly) four solar cycles may result in some mixing between the two derived signals. The QBO signal is small and confined to equatorial latitudes. In Fig. 4 the time-series of the annual average 30mbar height data is presented for 130"W, 20"N (because this point shows the *National Centers for Environmental Prediction/ National Center for Atmospheric Research. highest value in Fig. 3(a)), along with the solar activity index. This plot is similar to those presented before by Labitzke and Van Loon (e.g. 1997). The 1999 geopotential height value appears anomalous - it will be interesting to see if the curve turns up in response to the current near-solar-maximum conditions. Can the apparently solar-induced changes in Z3, be understood in terms of the observed changes in TSI? Using the radiative forcing arguments described in the previous section, the variation of 1.4Wm-2 observed between the 1 1-year cycle minimum in 1986 and maximum in 1991 can be deduced to potentially cause a mean increase in global average surface temperature of about 0.15 degc (neglecting any lag in the system). Analysis of sea surface temperature records by White et al. (1997) does indeed show variations of magnitude up to 0.14 degc on decadal times-scales with the largest response in mid latitudes. However, a warming of the entire atmosphere below 30mbar by 0.14degC would result in an increase in Z30 of only about 14m. Alternatively, to produce an increase in Z3, of loom would require a warming of 1 degc throughout the troposphere and lower stratosphere. Thus, if the pattern shown in Fig. 3(a) is a true response to solar variability, then some factors other than direct radiative forcing must be responsible. The geographical distribution of the variation in Z30 indicates an adjustment of the dynamical structure of the troposphere, and this is also suggested by GCM studies, but the magnitude of the apparent Z30 response, particularly in the summer hemisphere, has yet to be explained. Various factors which might ampllfy the solar effect have been proposed; two which have received particular attention involve (i) changes in stratospheric ozone and circulatory effects and (ii) changes in atmospheric ionisation by GCRs. These are briefly considered in the following sections. Solar ultraviolet and ozone The variability in TSI of order a few tenths of 1 %, discussed above, represents the changes integrated across the whole electromagnetic spectrum. In the ultraviolet (w), however, the amplitude of variability is much higher. Mea- 403

6 2385L Figure 4 Annual mean 30mbar geopotential height (m) at N (solid line), and 10.7cm solar radio (dashed line) (arbitrary units) surements of the solar w spectrum made by instruments on the Upper Atmosphere Research Satellite (Brueckner et al. 1993; Rottman et al. 1993), and also the results of theoretical models (Solanki and Unruh 1998), suggest an increase of -7% at nm and -3.5% at 250 nm from solar minimum to maximum. This variation in the spectrum complicates the issue of whereabouts in the atmosphere/ surface system the solar energy is deposited. Most of the visible and near-infrared radiation passes through the atmosphere unhindered to the tropopause and hence, neglecting scattering by cloud, to the surface, although water vapour bands in the near-infrared cause some absorption in the lower troposphere. Shorter LJV wavelengths, however, are absorbed in the middle atmosphere where they cause local heating and ozone production. The increased ozone tends to mask the lower atmosphere from the enhanced incident w while the warmer stratosphere will cause increased emission of thermal infrared (TIR) radiation into the troposphere (Haigh 1994). Thus the nature of the changes in the UV and TIR radiation fields depends on the ozone response. However, the linking of ozone to solar activity is not well established; two-dimensional photochemistry- 404 transport models of the stratosphere (e.g. Haigh 1994; Fleming et al. 1995) predict largest fractional changes in the middle-upper stratosphere with monotonically decreasing effects towards the tropopause, while multipleregression analyses of ozone measurements made from satellites (McCormack and Hood 1996; Wang et al. 1996) suggest largest changes in the upper stratosphere, zero, or even slightly negative, changes in the middle stratosphere, and significant positive changes in the lower stratosphere. However, as the data are only available over about one and a half solar cycles, have large uncertainties, especially in the lower stratosphere, and may not properly have accounted for the effects of volcanic aerosol (Solomon et al. 1996), the true nature of solarinduced changes in stratospheric ozone, and thus of the resultant changes in radiative forcing, remains uncertain. Changes in stratospheric thermal structure may also affect the troposphere through dynamica1 interactions rather than through radiative forcing. Kodera (1995) suggested that changes in stratospheric zonal wind structure, brought about by enhanced solar heating, could interact with vertically propagating planetary-scale waves in the winter hemisphere to produce a particular mode of response. This mode shows

7 dipole anomalies in zonal wind structure which propagate down, over the winter period, into the troposphere, and has been demonstrated in a GCM study by Shindell et al. (1999). Thus modifications to the stratosphere may result in modulation of tropospheric modes, such as the Arctic Oscillation. If this does take place, as seems possible but not yet proven, then there is scope for significant local variations in meteorological parameters in response to solar activity. Such a mechanism might contribute towards the apparently solar-induced changes in Z30 in the winter hemisphere but does not provide a simple explanation for the apparent summer hemisphere response. GCRS and atmospheric ionisation GCRs are particles which are formed outside the solar system and which bombard it from all directions. The flux of GCRs reaching the earth is modulated by interaction with magnetic structures advected with the solar wind such that at times of higher solar activity GCR flux is reduced by about 20% with respect to periods of lower activity. The flux into the atmosphere is also affected by the earth's magnetic field such that it is greatest at high latitudes. The GCRs which do penetrate the atmosphere are a major source of ionisation, particularly in the lower stratosphere. There are two main theories advanced as to the means whereby variations in GCR flux might impact climate. The first (Tinsley and Deen 199 1) concerns modulation of the earth's electric field; ionisation of the atmosphere by GCRs influences its conductivity and thus the current within the earth's electric circuit. It is plausible that the current flow into clouds may affect cloud microphysics, ice formation and hence thunderstorm activity, although details of this mechanism are not fully established. The second means whereby it is proposed that GCRs may affect climate (Dickinson 1975) is through enhancement of the production of cloud condensation nuclei through growth of aerosol on ionised air molecules. Evidence for the existence of this process has been obtained from observational and modelling studies, although not as yet in response to solar activity. Clearly, any changes in cloud cover, or albedo, have the potential to significantly impact the radiation balance. It has been demonstrated (Marsh and Svensmark 2000) that total cloud cover varied in phase with GCR flux between 1984 and 1995, but the response is confined to marine low cloud at low latitudes so that any link with GCR ionisation, which is greatest in high latitudes at higher altitudes, needs explanation. Conclusions It is important to understand how solar variability affects climate so that human and natural signals may be disentangled in the observational record and thus more reliable predictions made of the effects of human activity on future climate. However, the absolute value of the solar constant is not known to better than about 4W m-2 and reconstructions of past values are uncertain, even over the twentieth century and certainly over longer periods. Thus, longer-term and more accurate measurements are required before trends in TSI can be monitored to sufficient accuracy for direct application in studies of climate change. Analyses of global mean temperature records suggest a detectable signal of solar influence. Geographical variations of this influence are very poorly known but almost certainly do not consist of a uniform surface warming. Vertical patterns of temperature change may provide a method for differentiating solar from other climate perturbations. Variations in solar w radiation and solarinduced changes in stratospheric ozone probably have a small effect on radiative forcing but additionally may affect climate through a dynamica1 response to heating of the lower stratosphere. This heating depends partially on the ozone changes but the vertical structure of the ozone response is not well known. In the winter hemisphere changes in the thermal structure of the stratosphere may affect tropospheric climate by modlfylng planetary wave propagation but the apparent response to solar variability seen in 230 in the summer hemisphere is not well understood. The potential advantage of theories of solarclimate links which rely on GCRs over those based on electromagnetic radiation are two- 405

8 ~ (1996) ~ (1999) ~ (1997) ~ (1998) Weather Vol. 55 November 2000 fold. Firstly, they circumvent the argument that solar variations are energetically too small to produce the apparent effects. Secondly, the atmospheric response time is faster. However, much further research into the proposed mechanisms is required, alongside further careful observational studies, before their existence is established. Acknowledgements The multiple regression study was carried out at Imperial College by Jason Lam and Lalith Vipulananthan; the data were provided by the Stratospheric Research Group at FUB and a multiple regression program was provided by Myles Allen. I am grateful to Judith Lean for Fig. 1, to Ken Schatten, Judith Lean, Sami Solanki and Mike Lockwood for the sunspot and TSI data in Fig. 2 and to Mike Mann and the Carbon Dioxide Information Analysis Center for the temperature data. References Baliunas, S. and Jastrow, R. (1990) Evidence for long-term brightness changes of solar-type stars. Nature, 348, pp Bradley, R. S. and Jones, P. D. (1993) Little Ice Age summer temperature variations: Their nature and relevance to recent global warming trends. Holocene, 3, pp Brueckner, G. E., Edlow, K. L., Floyd, L. E., Lean, J. and VanHoosier, M. E. (1993) The Solar Ultraviolet Spectral Irradiance Monitor (SUSIM) experiment onboard the Upper Atmosphere Research Satellite (UARS). J. Geophys. Res., 98, pp Cubasch, U., Voss, R., Hegerl, G. C., Waszkewitz, J. and Crowley, T. J. (1997) Simulation of the influence of solar radiation variations on the global climate with an ocean-atmosphere general circulation model. Clim. Dyn., 13, pp Dickinson, R. E. (1975) Solar variability and the lower atmosphere. Bull. Am. Meteorol. SOC., 56, pp Fleming, E. L., Chandra, S., Jackman, C. H., Considine, D. B. and Douglas, A. R. (1995) The middle atmosphere response to short and long term uv variations: An analysis of observations and 2D model results. J. Amos. Err. Phys., 57, pp Frohlich, C. and Lean, J. (1998) The Sun s total irradiance: Cycles, trends and related climate 406 change uncertainties since Geophys. Res. Lett., 25, pp Gillett, N. I?, Allen, M. R. and Tett, S. F. B. (2000) Modelled and observed variability in atmospheric vertical temperature structure. Clim. Dyn., 16, pp Haigh, J. D. (1994) The role of stratospheric ozone in modulating the solar radiative forcing of climate. Nature, 370, pp The impact of solar variability on climate. Science, 212, pp A GCM study of climate change in response to the 11-year solar cycle. Q. J. R. Meteorol. SOC., 125, pp Herschel, W. (1801) Observations tending to investigate the nature of the Sun, in order to find the causes or symptoms of its variable emission of light and heat; with remarks on the use that may possibly be drawn from solar observations. Phil. Trans., 91, pp Hoyt, D. V. and Schatten, K. H. (1993) A discussion of plausible solar irradiance variations, J. Geophys. Res.,98,pp The role of the sun in climate change. Oxford University Press Group sunspot numbers: A new solar activity reconstruction. Solar Phys., 181, pp Intergovernmental Panel on Climate Change (1995) Climate change Cambridge University Press Jones, P. D., Parker, D. E., Osbom, T. J. and Briffa, K. R. (1999) Global and hemispheric temperature anomalies - land and marine instrumental records. In: Trends: A compendium of data on global change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, US Department of Energy, Oak Ridge, USA Kodera, K. (1995) On the origin and nature of the interannual variability of the winter stratospheric circulation in the northern hemisphere. J. Geophys. Res., 100, pp Labitzke, K. and Van Loon, H. (1997) The signal of the 11-year sunspot cycle in the upper troposphere-lower stratosphere. Space Sci. Rev., 80, pp Larkin, A., Haigh, J. D. and Djavidnia, S. (2000) The effect of solar w irradiance variations on the Earth s atmosphere. Space Sci. Rev. (In Press) Lean, J. and Rind, D. (1998) Climate forcing by changing solar radiation. J. Clim., 11, pp Lean, J., Skumanitch, A. and White, 0. (1992) Estimating the Sun s radiative output during the Maunder Minimum. Geophys. Res. Lett., 19, pp Lean, J., Beer, J. and Bradley, R. S. (1995) Reconstruction of solar irradiance since 1610: Implications for climate change. Geophys. Res. Lett., 22,

9 pp Lockwood, M. and Stamper, R. (1999) Long-term drift of the coronal source magnetic flux and total solar irradiance. Geophys. Res. Len., 26, pp McCormack, J. P. and Hood, L. L. (1996) Apparent solar cycle variations of upper stratospheric ozone and temperature: Latitudinal and seasonal dependences. J. Geophys. Res., 101. pp Mann, M. E., Bradley, R. S. and Hughes, M. K. (1998) Global-scale temperature patterns and climate forcing over the past six centuries. Nature, 392, pp Marsh, N. and Svensmark, H. (2000) Cosmic rays, clouds and climate. Space Sci. Rev. (In Press) Nesme-Ribes, E., Ferriera, E. N., Sadourny, R., LeTreut, H. and Li, Z. X. (1993) Solar dynamics and its impact on solar irradiance and the terresma1 climate. J. Geophys. Res., 98, pp Rind, D., Lean, J. and Healy, R. (1999) Simulated time-dependent climate response to solar radiative forcing since J. Geophys. Res., 104, pp Rottman, G. J., Woods, T. N. and Spam, T. P. (1993) Solar Stellar Irradiance Comparison Experiment: Instrument design and operation. J. Geophys. Res., 98, pp Sato, M., Hansen, J., McCormick, M. P. and Pollack, J. B. (1993) Stratospheric aerosol optical depths, J. Geophys. Res., 98, pp Shindell, D., Rind, D., Balachandran, N., Lean, J. and Lonergan, P. (1999) Solar cycle variability, ozone, and climate. Science, 284, pp Solanki, S. K. and Fligge, M. (1998) Solar irradiance since 1874 revisited. Geophys. Res. Lett., 25, pp Solanki, S. K. and Unruh, Y. C. (1998) A model of the wavelength dependence of solar irradiance variations. Astron. Astrophys., 329, pp Solomon, S., Portmann, R. W., Garcia, R. R., Thomason, L. W., Poole, L. R. and McCormick, M. P. (1996) The role of aerosol variations in anthropogenic ozone depletion at northern midlatitudes. J. Geophys. Res., 101, pp Stevens, M. J. and North, G. R. (1996) Detection of the climate response to the solar cycle. J. Atmos. Sci., 53, pp Tett, S. F. B., Stott, P. A., Allen, M. R., Ingram, W. J. and Mitchell, J. F. B. (1999) Causes of twentieth century temperature change near the Earth s surface. Nature, 399, pp Tinsley, B. A. and Deen, G. W. (1991) Apparent tropospheric response to MeV-GeV particle flux variations: A connection via electrofreezing of supercooled water in high-level clouds? J. Geophys. Res., 96, pp Van Loon, H. and Labitzke, K. (1998) The global range of the stratospheric decadal wave. Part I: Its association with the sunspot cycle in summer and in the annual mean, and with the troposphere. J. Clim., 11, pp Wang, H. J., Cunnold, D. M. and Bao, X. (1996) A critical analysis of stratospheric aerosol and gas experiment ozone trends. J. Geophys. Res., 101, pp White, W. B., Lean, J., Cayan, D. R. and Dettinger, M. D. (1997) Response of global upper ocean temperature to changing solar irradiance. J. Geophys. Res., 102, pp Willson, R. C. (1997) Total solar irradiance trend during solar cycles 21 and 22. Science, 277, pp Correspondence to: Dr J. D. Haigh, Department of Physics, Blacken Laboratory, Imperial College of Science, Technology and Medicine, London sw7 2sw. The Carrbridge cloudburst of 1923* David McConnell Bracknell, Berkshire This article is a sequel to the story of the Cambridge cloudburst of 1914 (McConnell2000). A more detailed account of this article, by David McConnell, appeared in the October issue of the monthly railway history magazine Backtrack. With the collapse of the Baddengorm railway bridge in 1914, the Highland Railway Company had used its original and more circuitous line from Aviemore to Inverness via Grantownon-Spey, Forres and Nairn, until a new bridge 407