Atmospheric effects on high energy cosmic rays

The atmospheric effects on the muon intensity have been observed using muon telescopes at Poatina, Cambridge and Hobart, Tasmania. The detectors consist of a variety of Geiger counter telescopes located at ‚ÄövÑvÆ 357 hg cm-2 underground, ~ 47 hg cm-2 underground and at sea level, respectively. Least-squares regression analysis has been used to calculate the various pressure, height and temperature coefficients for the three sites. It is found that for the underground detectors, the apparent atmospheric effects tend to differ strongly from those expected. In particular, the total barometer coefficient at Poatina for the years 1972-76 is (-0.047 ¬¨¬± 0.002) % mb-1 (Humble et al., 1979). Using a generally-accepted model for the production and interaction of secondary particles in the atmosphere, extensive calculations predict a barometer coefficient of only ~ 0.003 % mb-1. The partial pressure, height and temperature coefficients also differ greatly from their theoretical values. A detailed review is presented of the theory and assumptions behind the least-squares method. It is shown analytically that if the regressor variables are inter-correlated and measured with error, their expectation values may be severely biased from their true values. It is found that the inflated Poatina total barometer coefficient is almost entirely due to a negative correlation between air pressure and stratospheric temperature. The latter variable is thus shown to be the only significant influence on the intensity of muons deep underground. A comparison between the observed and predicted total temperature coefficients implies a temperature measurement error of ~ 2.2 K. This error is partly instrumental and partly due to the slow sampling rate of temperature provided by radiosondes. In all, these data problems are probably entirely responsible for the anomalous coefficients at Poatina. A test parameter, based on the least-squares model, is shown to predict successfully the observed divergence between the observed and theoretical coefficients, as the muon threshold energy is increased. An extensive series of regressions has been carried out in order to determine the optimal set of atmospheric variables for correcting the Poatina intensity for atmospheric effects. However, the percentage of variation removed is still only ~ 40 %, because of the data intercorrelations and errors previously mentioned. Using the temperature dependence of the Poatina muon intensity, the four-day mean temperature of the stratosphere has been reconstructed. The resulting correlation with radio-sonde measurements is ~ 0.5. However, a variety of detectors with greater sensitive area would be required for the determination of the atmosphere's temperature profile from muon observations to be practicable.