Data and code for: Conditional and probabilistic attribution of Japan's August–September 2026: robust heat, signal-dependent storyline rainfall

Data and code supporting the article “Conditional and probabilistic attribution of Japan's August–September 2026: robust heat, signal-dependent storyline rainfall” by Tomohiro Akiyama. The study compares a circulation-conditioned storyline (five pairs of spectrally nudged 9-km WRF simulations with and without the anthropogenic thermodynamic signal) with probabilistic attribution based on observations, CMIP6 and d4PDF, for 17 seasonal quantities of August–September 2026 in Japan. Both methods detect the heat; the storyline rainfall decreases depend on how the imposed signal is constructed. The archive contains the derived results of both methods and of their comparison, the post hoc checks (signal uncertainty, likelihood ratios, scaling frames, the family of sensitivity tests, denominators and lateral moisture inflow), the replicated uniform-signal pairs and the same-input platform comparison of one pair, the analysis scripts and the scripts and settings kept on the computers, the frozen protocols and pre-specifications with their fingerprints, the figures, the study's rows of the run ledger, records of the software environments, rights ledgers and verification tools (QUICK_VERIFY.py). Original ERA5, CMIP6, d4PDF and JMA downloads and the WRF output are not included; INPUTS_NOT_INCLUDED.tsv lists them with their routes and fingerprints. Licences: code under the MIT License; author-created data, results, figures and documents under CC BY 4.0; third-party terms in NOTICE.md. This research was supported by Uehiro Research Center for Japan Environmental Studies (Endowed Research Division) of Research Institute for Humanity and Nature, National Institutes for the Humanities.