Short-block tests did not show a clear cold-exposure benefit from cherry bloom timing at three sites in 1991-2024

We joined annual cherry bloom records at Kyoto, Liestal and Washington to daily minimum air temperature on a 0.5 degree grid. We tested 1991-2007 and 2008-2024 separately. We compared the seven-day minimum from bloom with a reference that assigns dates to weather years within four- or five-year blocks. Actual-year minima were 0.27 to 0.91 C higher than the reference in all six tests. No test had a two-sided permutation p value below 0.05, before or after six-test adjustment. Wider year shuffles gave stronger results at Washington. Some results also changed with window length, date shifts and grid cell. Thus, these tests did not establish a common cold-exposure benefit from annual bloom timing. This is a test of annual weather and timing, not a measure of damage or a causal climate effect. Fixed sites and complete years reduce changes in site coverage. Different bloom rules, unknown visit records, coarse weather and only 17 years per test limit inference. The package includes exposure matrices, source hashes, code and checks.

phenology · cherry bloom · cold exposure · observation bias · permutation test · temporal replication · null result