Derived verification data for gauge-referenced point nowcasting
收藏资源简介:
Verification data supporting "Gauge-referenced hourly rainfall nowcasting: Comparing direct radar-to-point and field-first forecast routes". Supplementary_Data_1.xlsx is identical to the current journal-submission workbook. It contains a README and 59 data/provenance worksheets ordered by manuscript figures, followed by supplementary figures and training metadata. Table 1 values are adjacent to Figure 4 data, case definitions accompany Figure 7, and station metadata accompany Figure 8. The workbook provides observation-defined rainfall-episode peak-amount and timing verification, duration statistics, and exPreCast CNN 3 × 3 versus 5 × 5 patch-size sensitivity. Figures 6–10 cover episode errors, cases, stationwise skill, distance-group comparisons and gauge-input/dropout diagnostics; Supplementary Figures S2–S3 cover patch sensitivity and episode duration. Paired patch-size CSI differences and confidence intervals are provided in FigS2_paired_CI. The Figure 5 sheets provide gauge-conditioned amount biases by observed accumulation bin, including route means, between-member sample standard deviations and common-pair counts, together with exceedance-frequency diagnostics. CNN_selected_epochs records source filenames and SHA-256 hashes for the training records used in epoch selection. The workbook includes station metadata, selected case-study gauge observations and forecast series, and observation-derived episode summaries. The complete raw KMA radar and gauge archives, full prediction arrays, model weights and credentials are not redistributed. Retain the KMA attribution and source terms in README.md. The corresponding software is r2p-nowcasting version 1.2.1, distributed separately. Correspondence: Joon-Woo Roh (neo@earthntech.co.kr). Minor file correction (2026-09-29): author affiliation and corresponding email were corrected, with associated checksums updated. Only the README and its checksum changed; Supplementary_Data_1.xlsx is unchanged.



