A method for quantifying the detector systematic uncertainties of a large water Cherenkov far detector (Super-Kamiokande in the T2K oscillation experiment), where reconstruction mis-modeling is one of the leading uncertainties on the extracted oscillation parameters. Rather than modeling each physical effect (light scattering in water, photomultiplier response) separately, the method is deliberately source-agnostic: any mis-modeling is assumed to appear at analysis level as a distortion of the log-likelihood particle-identification (PID) variables that separate electron-like from muon-like events, and is captured by a two-parameter shift-and-smear transform of the simulated PID value, L -> alpha*L + beta. The parameters are inferred by fitting simulated Monte Carlo to real data on a high-statistics control sample - decades of Super-Kamiokande atmospheric-neutrino events - inside a Markov Chain Monte Carlo that proposes alpha/beta values and scores a shape likelihood, yielding a posterior. Uncertainties are separated by true event topology above Cherenkov threshold so they propagate correctly to the beam experiment's distinct signal and background samples, and the atmospheric-derived systematics are transferred to the beam oscillation analysis. The contribution is this agnostic, control-sample-anchored, MCMC-based detector-uncertainty estimation chain, upgraded to a 540-dimensional fit for new analysis samples.
This conference-proceedings contribution (NuFact 2024) describes how the T2K long-baseline neutrino oscillation experiment estimates the detector systematic uncertainties of its far detector, Super-K…