Superior and Michigan-Huron show no clear low-high recovery time difference at a five-year limit for 1918-2020 start years

We tested whether low annual water levels took longer than high levels to reach the historical median in Superior and Michigan-Huron. We used coordinated v2025 monthly data for 1918-2025. Low and high start years were in the lower and upper quarters of annual means. We compared first-passage times limited to five years, for start years through 2020. The low-minus-high mean was 0.81 years in Superior, with a conditional 95% ten-year block interval of -0.39 to 1.94 years. It was 0.59 years in Michigan-Huron, with limits of -0.30 to 1.69. All 16 planned intervals included zero. The estimates suggest longer waits from low states, but the record does not resolve either difference. Only 6-9 separate tail runs support each primary group. Overlapping start years, a five-year cap, fixed historical thresholds, unequal tail distances, possible changes over time and one product limit inference. This is a historical state-based comparison, not an episode-onset estimate, physical cause, or future risk model. The package includes input rows, source hashes, all outcomes, bootstrap draws, code, a standard-library audit and a figure.

Great Lakes · water levels · recovery time · first passage · persistence · null result