But quick example: our 38 km^2 headwater basin produced a higher routed peak under the 6-hr PMP than the 24-hr, which ended up controlling spillway crest and freeboard to meet overtopping risk targets — anyone else see short-duration PMP govern? We confirmed with HEC-HMS and a 2.5-hr Tc; curious if this aligns with your unit hydrograph work.
Same on our about 30 km² headwaters: with a ‘2.5-hr Tc’, the 6-hr PMP hammers the Clark UH and ends up governing crest, not the 24-hr. If you haven’t already, flip the HEC-HMS temporal pattern to front- or center-loaded and test Atlas 14 distributions for burst sensitivity: PF Data Server-PFDS/HDSC/OWP. Caveat: once you add meaningful detention or channel storage, the 24-hr can retake freeboard.
With a ‘2.5-hr Tc’ I’ve seen the 6‑hr PMP govern, but only after I cut the HEC‑HMS compute timestep to 5–10 min and used ModClark so the front‑loaded pattern isn’t smeared… @OP small caveat: if your reservoir routing time is long, the 24‑hr can retake control. What timestep and Clark R are you running?
I’ve had the short‑duration PMP govern too; the lever was sliding the temporal pattern so the biggest 1–2 h block coincides with your about 2.5 h concentration time — routed peak jumped enough to set crest/freeboard. Small caveat: if you’ve got appreciable reach detention, the longer storm can still control. @OP did you try a simple ‘temporal shift’ sweep of the pattern start?