Geert Barentsen, Rainer Arlt, Hans-Erich Fröhlich
A method for estimating the true meteor rate \lambda\ from a small number of
observed meteors n is derived. We employ Bayesian inference with a Poissonian
likelihood function. We discuss the choice of a suitable prior and propose the
adoption of Jeffreys prior, P(\lambda)=\lambda^{-0.5}, which yields an
expectation value E(\lambda) = n+0.5 for any n \geq 0. We update the ZHR meteor
activity formula accordingly, and explain how 68%- and 95%-confidence intervals
can be computed.
View original:
http://arxiv.org/abs/1112.4372
No comments:
Post a Comment