Has anyone validated Eckhardt vs Lyne-Hollick for baseflow separation on 15‑min discharge in small urban watersheds? I’m analyzing 2018–2024 USGS 15‑min flows at gage 06730500 using R (EcoHydRology), calibrating alpha and BFImax via split-sample NSE on log(Q) and a master recession curve, but the parameters look unstable across seasons. My plan is block bootstrap with season-stratified cross-validation and to report uncertainty bands — does that sound defensible, or would you fix parameters purely from recession analysis; and for anyone starting this, document your priors and align the objective function with the behavior you care about (e.g., low-flow bias)?