Rainfall-product choice alters normalized storage-reliability estimates in Ambovombe, southern Madagascar: Reproduction package

This record archives the complete computational reproduction package for the study “Rainfall-product choice alters normalized storage-reliability estimates in Ambovombe, southern Madagascar.” Study scope The analysis covers Ambovombe commune in the Androy Region of southern Madagascar. The dissolved study polygon covers 477.8 km². Three daily rainfall products were analyzed over the common period from January 1, 1995, through December 31, 2024, comprising 10,958 consecutive days: - Direction Générale de la Météorologie Rainfall Estimate, DGM RFE;- Climate Hazards Infrared Precipitation with Stations, CHIRPS2 Final;- ERA5-Land. The products were treated as equal-rank rainfall representations. No product was designated as an observation or accuracy reference because no continuous local daily gauge record covering the complete study period was available for direct evaluation. Scientific purpose The package supports the calculation and attribution of differences in normalized storage-reliability thresholds among the three rainfall products. The analysis separates: - differences in long-term mean effective rainfall;- differences in monthly input trajectories;- differences in the daily allocation of rainfall within months. The attribution combines product-wise mean scaling, a 3 × 3 recombination of monthly totals and intramonth daily profiles, 1,999 paired intramonth randomizations, and conditional factorial decomposition. Package contents - SM1 provides the detailed methods, equations and references.- SM2 contains the exact executed analysis, validated daily rainfall series, analysis script, configuration, intermediate outputs, logs, manifests and quality-control results.- SM3 contains eight publication graphics, plotting data, captions and the figure-generation script.- SM4 contains six manuscript tables, machine-readable table data and the table-generation script.- SM5 contains the three QGIS rainfall collectors and the Ambovombe study polygon.- SM6 provides portable execution paths, dependency versions, provenance records, expected results, geographic files, licensing information and independent verification scripts. Reproduction Two reproduction paths are documented: 1. exact reproduction from the three validated polygon-level daily series retained in SM2;2. recollection of DGM RFE, CHIRPS2 Final and ERA5-Land from their respective public source services using the SM5 collectors and the archived study polygon. The three canonical analysis inputs use the schema: "date, unit_code, rainfall_mm, rainfall_volume_m3" Archived execution checks The completed execution passed: - 28 mandatory synthetic tests;- 204 quality-control checks;- zero warnings;- zero failed checks. The central time-reliability storage thresholds were 65.0, 67.5 and 122.5 days of demand for DGM RFE, CHIRPS and ERA5-Land, respectively. The corresponding volume-reliability thresholds were 62.5, 65.0 and 120.0 days. Monthly-total source accounted for approximately 99.7% of the observed time-threshold variation and 99.5% of the observed volume-threshold variation. Its sum of squares exceeded those of daily-profile source and interaction in all 1,999 generated daily orders for both reliability metrics. These values are normalized diagnostic thresholds and are not direct dimensions for a physical reservoir. Data and licensing The polygon-aggregated validated daily analytical series are retained in SM2 for exact reproduction. Native provider grids, provider caches, service credentials and authentication tokens are not included. Original code is licensed under the MIT License. Original documentation, the study geometry and author-created derived outputs are licensed under Creative Commons Attribution 4.0 International. Third-party rainfall and boundary products remain attributed to their respective providers. Package DOI: 10.5281/zenodo.21531316