Random year order matched winter precipitation transition rates in seven California divisions in 1901-1960 and 1966-2025

We tested whether changed year ordering is needed to explain the difference in winter precipitation transitions between two 60-year periods in seven California divisions. We used fixed NOAA nClimDiv data and early-period 20th and 80th percentile thresholds. The observed rate of transitions between opposite tails was 8.23% in 1901-1960 and 13.32% in 1966-2025. The change was 5.08 percentage points, with a conditional 95% five-pair block interval of -1.69 to 11.62 points. Random year order with the observed tail counts gives rates of 8.14% and 13.30%. Its expected change of 5.17 points closely matches the observed difference. The change in the departure from that reference was -0.08 points, with a two-sided permutation p value of 1.000. Common permutations kept all seven divisions together. Separate percentile thresholds for each era removed the observed increase; a shorter-period comparison gave no change. These results do not prove independence or establish a trend. They show that the observed difference does not require an extra ordering effect under the tested null. Short records, threshold choice, exchangeability assumptions, and one retrospective product limit inference. The package includes exact source rows, code, all draws, figures, an independent audit, and an exact same-host repeat.

California · precipitation · winter · whiplash · opposite-tail transitions · temporal ordering · conditional permutation · nClimDiv · null result · spatial dependence