No-flow days at 137 US drying stream gauges did not rise past their 1999-2017 level in 2018-2025, except in western deserts and mountains; P/PET did not explain the shifts

Zipper et al. (2021) found that annual no-flow days at US non-perennial stream gauges changed in 1980-2017: drying in the south and west and wetting in the north. We tested if these changes continued in eight new climate years, 2018-2025. We used USGS NWIS daily discharge for 9,067 GAGES-II gauges and the same no-flow rule. We reproduced the published Mann-Kendall trend class for 96.8% of 539 matched gauges. Of 541 gauges with test data, 137 had a drying trend. In 2018-2025, 74% of these stayed above their 1980-2017 mean, but only 52% rose past their 1999-2017 mean (p = 0.73; median +1.5 days per year). In a placebo split inside the published period, 68% of drying gauges continued past their recent level in 2010-2017 (p = 0.002), so the stall is new. Wetting gauges continued a little (67% lower than their 1999-2017 mean, p = 0.002). Drying continued in the Western Deserts (13 of 16 gauges) and less in the Western Mountains. The 1999-2017 mean gave the lowest mean absolute error (29.0 days per year). Sen-trend extension (32.5) and a per-gauge model with basin precipitation over Thornthwaite PET (33.1) were worse. The climate model explained a median 33% of yearly variance. After the climate adjustment, 46% of drying trends and 70% of wetting trends stayed significant, and in 2018-2025 the trend gauges kept their offsets from the climate forecast. Thus, the changes acted as persistent level shifts that current- and prior-year P/PET did not explain. Limitations: only eight test years, near-zero discharge measurement sensitivity, most gauges are not reference gauges, temperature-only PET, and spatial clustering (HUC-2 cluster intervals include zero for some forecast differences). Code, derived public-domain tables, figures and source hashes are included. The CC BY-NC-SA Zipper et al. release is not bundled.

stream intermittency · no-flow days · non-perennial streams · out-of-sample test · trend persistence · USGS NWIS · GAGES-II · nClimGrid · aridity · replication · drought · hydrology