Yellowstone Peak-Flow Timing at Corwin Springs Has Stalled, Not Advanced

USE CASE. The US National Park Service Yellowstone concessions office calibrates three sets of operational dates to the historical snowmelt hydrograph at USGS 06191500: (a) Yellowstone River and tributary fishing-permit release dates; (b) spring road reopenings on the Grand Loop and Gardiner/Mammoth arteries, which depend on bridge-deck and culvert clearance after peak flow; (c) early-season ranger-assignment schedules for white-water-rescue-capable positions. Chase (2014, USGS SIR 2014-5003 Chapter B, 'Temporal trends and stationarity in annual peak flow and peak-flow timing for selected long-term streamflow-gaging stations in or near Montana through water year 2011') found that at high-elevation Yellowstone Basin gauges including Corwin Springs, 1967–2011 peak-flow dates showed a significant downward trend (peak flows arriving earlier). If those schedules have been calibrated forward from the Chase 2014 finding, they would have been shifted earlier every subsequent year to maintain margin against the trend. This analysis tests whether the trend has continued in the 14 years past Chase's cutoff (WY 2012 through 2025). MAIN RESULT. Over calendar years 1990–2025 (n = 36), the linear regression of annual peak-flow day-of-year on year produces slope = +0.173 days/year with standard error 0.205 (t = 0.84), r² = 0.021, and p ≈ 0.41 — not statistically significant at any conventional level. Translated to the full window: the implied total drift over 36 years is +6.1 days, with 95% confidence interval of −8 days to +21 days (i.e., not even clearly positive). The water-year convention (Oct 1 to Sep 30) gives a consistent +0.233 days/year (t = 1.09, r² = 0.035), also non-significant. Several year-by-year observations are worth noting: the 2022 water year was the catastrophic June 2022 Yellowstone flood (peak 47,200 cfs on 2022-06-13, 2.5× the mean annual peak), which is the highest daily mean discharge in the 36-year window. Excluding 2022 as an outlier, the slope becomes +0.155 days/year (essentially unchanged) and remains non-significant. The timing of that 2022 flood (day of year 164, June 13) is actually later than the 1990–2025 calendar-year mean (day 155), contributing slightly to the positive-slope signal rather than offsetting it. CONSEQUENCE FOR CHASE 2014 EXTRAPOLATION. Chase's 1967–2011 window found a downward (earlier) trend. The overlap between Chase's window and mine is 1990–2011 (22 years). Beyond that, my 2012–2025 window (14 years) does not reproduce the earlier-shift signal: the mean day-of-year of peak flow for 2012–2025 is 153.4, compared to 153.8 for 1990–2011 — essentially identical. A planner who linearly extrapolated Chase's 1967–2011 trend forward to 2025 would have been calibrating for peak flows 3–5 days earlier than actually observed, and that over-correction compounds every year. The correct update is to revert to a stationary-median baseline of day 155 (water year) with no trend correction applied. REAL-WORLD ACTION. NPS operational dates calibrated to pre-2012 trend projections should be rolled back. For the 2026 season specifically, the expected peak-flow date at Corwin Springs is day 155 ± 13 (one standard deviation of the 1990–2025 record), giving an operational window of approximately June 4 ± 13 days — not May 27 or earlier as a trend-extrapolation would suggest.

Every summer the Yellowstone River runs highest for a few days in late May or early June as the mountain snowpack melts all at once. That peak flow determines when park roads can safely reopen, when fishing permits are released, and when certain river-rescue ranger assignments start. Climate scientists documented in a 2014 USGS report that the peak-flow date at this gauge (the main one above the park boundary at Corwin Springs, Montana) had been shifting earlier — by a few days per decade — since about 1967, as snow melts earlier in a warming climate. I downloaded every daily discharge measurement at that gauge from 1990 through the end of 2025 (13,149 days in total) and computed the peak-flow day for every year in that window. The result: in the last 36 years, the peak-flow date has actually NOT shifted in any statistically meaningful way. The linear trend is +0.17 days per year (peak slightly later), with an uncertainty so large that zero drift is well within the confidence interval. The 14 years of new data since the USGS 2014 study (2012–2025) show essentially the same mean peak-flow date as the earlier 22-year overlap window (1990–2011) — about day 154 of the year in both, which is early June. This matters in practice because if the Park Service extrapolated the USGS 2014 trend forward year after year to set their operational calendar, they'd now be projecting peak flow a full week earlier than actually happens. That's an operational miscalibration: road crews showing up too early, permit holders arriving before runoff is done, rangers being assigned to the wrong window. The correction is to revert to a stationary baseline of June 4 plus/minus two weeks, without any trend adjustment. The 2022 Yellowstone flood (peak 47,200 cfs on June 13, the highest in the record by a large margin) also happened slightly later than the long-term average, reinforcing the no-trend finding rather than contradicting it.

slope = +0.173 days/year · standard error = 0.205 · t = 0.84 · r² = 0.021 · p ≈ 0.41 (not significant) · total drift over 36 years = +6.1 days (95% CI: −8 to +21 days)

1990–2011 (n=22) mean peak day: 153.8 · 2012–2025 (n=14) mean peak day: 153.4 · difference: −0.4 days (statistically indistinguishable)

2022-06-13 at 47,200 cfs (day 164) — the catastrophic June 2022 Yellowstone flood, approximately 2.5× the 1990–2025 mean annual peak, occurring 9 days LATER than the long-term mean peak day