Bloom and weather alignment gave no clear freezing-exposure reduction at two cherry sites in 1951-2025

We tested whether the annual pairing of cherry bloom dates and weather reduced freezing exposure at Liestal and Washington in 1951-1987 and 1988-2025. We used fixed bloom records and quality-approved GHCN daily minimum temperatures. Dates were exchanged within five-year calendar bins. Observed 14-day freezing exposure was 5.29, 9.81, 5.14 and 10.35 percentage points lower than the exchange expectation in the four tests. No exact permutation test passed correction for four tests; adjusted p values were 1.000, 0.460, 1.000 and 0.159. All four main bin-bootstrap intervals included zero. Planned window, threshold and bin-size checks also failed the adjusted permutation tests. The common direction is consistent with annual weather tracking but does not establish protection from frost. Exchangeability, few bins, station mismatch, missing weather and unknown observation changes limit inference. The test does not measure frost damage, pollution, biodiversity loss or a causal climate effect. Code, source hashes, binary exposure matrices, results and independent enumeration checks are included.

phenology · cherry bloom · freezing exposure · weather extremes · observation bias · permutation test · null result