Larger 3-day rain maxima did not give larger annual floods in 368 US rain-dominated basins in 2003-2025; lower pre-storm streamflow explains about half of the shortfall

Question: in US CAMELS reference basins, did annual maximum daily flow (AMF) follow the increase in heavy rain after 2002, and does lower antecedent wetness explain any shortfall (Sharma, Wasko and Lettenmaier 2018)? Approach: USGS daily discharge and Daymet v4 basin precipitation (3,454 sample points) for water years 1981-2025. Per basin, a storm-response model was fit on 1981-2002 declustered 3-day storms (4 per year) and tested on 2003-2025, with and without pre-storm streamflow and 30-day rain. Results (368 rain-dominated basins, 95% water-year bootstrap intervals, log units): the largest 3-day rain increased by 0.062 (0.014 to 0.105), but AMF did not increase (-0.056; -0.203 to 0.100). At fixed storm size the flood peak was 0.106 lower (-0.203 to 0.005); with wetness terms the decrease was 0.057 (-0.096 to -0.017). A rain-only model over-predicted the AMF change by 0.171 (0.091 to 0.256); with wetness terms, by 0.091 (0.031 to 0.166). A placebo split (1981-1991 against 1992-2002) gave -0.058 (-0.169 to 0.056). Most of the response decrease is in Plains-Texas and western (mainly California) rain basins; eastern and Midwest basins show none. The lower pre-storm flow is not explained by 3 years of prior rain and only slightly by warmer weather. Limitations: daily peaks, one precipitation product (its RX3 trend is about 0.016 larger than the CAMELS v1.2 Daymet trend), wetness proxies instead of soil moisture, statistical models, and wide regional intervals.

floods · extreme precipitation · antecedent wetness · CAMELS · streamflow · Daymet · USGS NWIS · out-of-sample test · baseflow · flood trends · rainfall-runoff response