With a 24-month cap, recovery from low 12-month precipitation ranks took longer than in reordered records across 344 US divisions, 1901-2025

We tested recovery from low precipitation ranks in 344 contiguous US climate divisions. We used NOAA nClimDiv monthly data for 1901-2025. An episode starts at the 10th percentile of a rolling 12-month total. It ends at the calendar-month median. We capped recovery time at 24 months and gave each division equal weight. The observed mean was 14.21 months. The mean from 1,999 reference records was 13.30 months, with a central 95% range of 12.91 to 13.73. The excess was 0.91 months. The upper-tail reference fraction was 0.0005. The reference reordered years separately for each calendar month within 25-year periods. Each monthly spatial field stayed together. Thresholds were fit again in each draw. All planned checks are reported. Long recovery thus occurred under the reference, but observed order added a small departure. The test measures precipitation ranks. It does not measure water-store recovery, find a physical cause, or test forecast skill. One source, fitted thresholds, the cap, and within-period exchangeability limit inference. The package contains all selected monthly values, event records, exact permutations, code, provenance, and an independent arithmetic audit.

precipitation · drought recovery · nClimDiv · moving sums · seasonality · reference model · spatial dependence