We propose a {em unitarity method} for determining $tau$ neutrino appearance probability $P( u_{mu} rightarrow u_{tau})$ in long-baseline (LBL) accelerator experiments and atmospheric neutrino observations. When simultaneous in situ measurements of $P( u_{mu} rightarrow u_{mu})$ and $P( u_{mu} rightarrow u_{e})$ proceed, as is typical in the LBL experiments, one can use unitarity to measure $P( u_{mu} rightarrow u_{tau})$. A theorists toy analysis for the model-independent determination of $P( u_{mu} rightarrow u_{mu})$ and $P( u_{mu} rightarrow u_{e})$ is presented by using the NOvA data. It is shown in our analysis that $lsim$5% (8%) measurement of $tau$ neutrino appearance probability in neutrino (antineutrino) mode is possible in the peak region $1.5 lesssim E_ u lesssim 2.5$ GeV. The $ u$SM-independent nature of determination of the probabilities is emphasized.
Download