The most controversial suggestion for a direct effect of cosmic rays on climate is that they directly modulate the formation of clouds (Friis-Christensen and Svensmark, 1997; Svensmark, 1998; Marsh and Svensmark, 2000, 2003; Udelhofen and Cess, 2001; Kristjansson and Kristiansen, 2000; Carslaw etal., 2002; Arnold and Neubert, 2002). This idea was first proposed by analogy to cloud chamber particle detectors - an analogy that is not valid because natural atmospheric supersaturations are much smaller than those needed to make a cloud chamber work. The idea has been revived by observed correlations over recent solar cycles between GCRs counts and the global composite of satellite cloud cover observations compiled by the International Satellite Cloud Climatology Project, ISCCP (Rossow et al., 1996). Udelhofen and Cess (2001) found a solar cycle signal in ground-based data from 90 weather stations across the North American Continent. Instrument relocation and changes mean that a long-term drift in these ground-based data cannot be determined, but de-trended data show a clear and persistent solar cycle variation in coastal cloud cover in data that extends back to 1900. The best correlations between GCRs and global cloud cover have been obtained by Marsh and Svensmark (2000) from the infrared observations of clouds (10-12 /m) that make up the "D2" set compiled by ISCCP: these authors find that it is primarily liquid, maritime clouds, away from regions of an El Nino (ENSO) event, that correlate well with GCR fluxes. Other authors argue that the results are still influenced by ENSO events (Farrar,
2000). Correlations on shorter timescales due to Forbush decreases in GCR fluxes, have been reported in localised datasets by Veretenenko and Pudovkin (1997). The arguments against such direct cosmic ray-cloud connection have been:
1. Given the atmospheric supersaturations, there is no known mechanism that can cause the effect.
2. The inter-calibrations involved in the composite ISCCP dataset render it unsuitable for this type of analysis.
3. The data sequences are too short and so the significance of the correlations is low or marginal.
4. Periods of low geomagnetic field, particularly the Laschamp event, gave enhanced GCR fluxes in the Earth's atmosphere but did not influence climate in the Greenland area (Beer, 2001).
Of these objections, only 4 argues that such a mechanism is not effective: 1-3 all argue that the evidence in its favour is inadequate but cannot be used as arguments that it is not active. In relation to 4, there is some evidence that the failure to see an effect of the Laschamp event may have been a local characteristic of the climate in the Greenland area (Christla et al., 2004). Nevertheless, objection 4 remains the key debate.
Concerning the potential mechanisms (objection 1), there are now some viable proposals. Air ions are produced in the atmosphere, broadly in proportion to the flux of incident GCRs (Aplin et al., 2005) and there are plausible mechanisms now known, by which these can grow into cloud condensation nuclei (CCNs) at natural atmospheric supersaturations (Carslaw et al., 2002). These include the growth of air ions into aerosols (Yu and Turco, 2000; Eichkorn et al., 2002) and the effects of atmospheric electric fields on the microphysical connection between aerosols and clouds (Tinsley et al., 2000; Tripathi and Harrison, 2002). Given these mechanism appear to be viable, the major debate now is if such an effect would be significant compared to the many other sources of CCNs (Carslaw et al., 2002).
In relation to objections 2 and 3, recent studies using longer series of homogeneous data are providing new and strong evidence that there could indeed be a direct cosmic ray-cloud effect in certain regions (Harrison and Stephenson, 2005). Specifically, the "diffuse fraction" (the ratio of the intensity of daylight under direct illumination and in the shadow of an obscuring shield) is a simple and repeatable measure of local cloudiness and aerosols that has been made at many weather stations for many years. Figure 3 summarises some of these results from stations in the UK. Part (a) shows that the diffuse fraction is 2% lower on the 14% of days when the Climax GCR count is less than 3.6 x 105 hr-1, compared to the days when it exceeds this threshold. This result is found at all the stations studied and is highly statistically significant (usually at or exceeding the 99.9% level) in almost all cases. Note that it is possible, or even probable, that the effect is only for stations in clean, maritime air (Wilding and Harrison, 2005) and appears to be less significant in regions of high rainfall.
Was this article helpful?