On the statistical significance of the GZK feature in the spectrum of ultra high energy cosmic rays

The nature of the unknown sources of ultra-high energy cosmic rays can be revealed through the detection of the GZK feature in the cosmic ray spectrum, resulting from the production of pions by ultra-high energy protons scattering off the cosmic microwave background. Here we show that the GZK feature cannot be accurately determined with the small sample of events with energies $\sim 10^{20}$ eV detected thus far by the largest two experiments, AGASA and HiRes. With the help of numerical simulations for the propagation of cosmic rays, we find the error bars around the GZK feature are dominated by fluctuations which leave a determination of the GZK feature unattainable at present. In addition, differing results from AGASA and HiRes suggest the presence of $\sim 30%$ systematic errors in the relative energy determination of the two experiments. Correcting for these systematics, the two experiments are brought into agreement at energies below $\sim 10^{20}$ eV. After simulating the GZK feature for many realizations and different injection spectra, we determine the best fit injection spectrum required to explain the observed spectra at energies above $10^{18.5}$ eV. We show that the discrepancy between the two experiments at the highest energies has low statistical significance (at the 2 $\sigma$ level) and that the corrected spectra are best fit by an injection spectrum with spectral index 2.5-2.6. Our results clearly show the need for much larger experiments such as Auger, EUSO, and OWL, that can increase the number of detected events by 2 orders of magnitude. Only large statistics experiments can finally prove or disprove the existence of the GZK feature in the cosmic ray spectrum.