MA-RAIN v1.0.1: Consistent Monthly Rainfall Dataset for Maranhão State, Brazil (1987–2023)

Description MA-RAIN v1.0.1 is a consistent, quality-controlled, and gap-filled monthly precipitation dataset for Maranhão State, Brazil, covering the period January 1987 to December 2023 (37 years; 444 months). The dataset underpins the findings reported in the peer-reviewed article published in Climate (MDPI): Reschke, G.A., Dias, C.W.S., Menezes, R.H.N., Chagas, F.P. & Silva-Junior, C.H.L. (2026). Filling the Gaps: Creating a Consistent Rainfall Dataset for Maranhão State, Brazil (1987–2023). Climate, 14(3), 63. https://doi.org/10.3390/cli14030063 The dataset was developed to address the pervasive challenges of incomplete and inconsistent long-term rainfall records in tropical regions — including missing values, irregular sampling, inter-station inhomogeneities, and network discontinuities — which critically limit the reliability of climatological analyses, trend detection, and hydrological modelling in data-sparse environments. The dataset integrates observations from 100 rain gauges and meteorological stations distributed across Maranhão and its immediate border areas, including stations in Pará, Tocantins, and Piauí used for spatial interpolation. Data were sourced from two national monitoring networks: the Brazilian National Water and Basic Sanitation Agency (ANA; 90 stations) and the National Institute of Meteorology (INMET; 10 stations). Stations were organised into ten Homogeneous Precipitation Regions (HPR1–HPR10) through multivariate analysis (Principal Component Analysis combined with cluster analysis), following the spatial subdivision proposed by Menezes (2009) for Maranhão, to ensure methodological consistency in gap-filling and regional analysis. Precipitation records underwent a rigorous three-stage processing workflow: Quality Control — Detection and removal of outliers, physically implausible values, and duplicated records. Monthly series for all 100 stations were individually inspected against regional benchmarks and long-term climatological expectations. Gap-Filling (Regional Weighting Method) — Of the 100 stations evaluated, 26 presented no missing months; the remaining 74 stations contained a total of 314 missing monthly records, with a maximum of 14 missing months per station. Gaps were estimated using the Regional Weighting Method (Equation 1 of the reference paper), which employs data from the three nearest neighbouring stations within the same HPR as references, weighting each neighbour's observed value by the ratio of that neighbour's observed value to its long-term monthly mean. Estimates were restricted to stations within the same HPR (maximum inter-station distance: 137.79 km), consistent with WMO recommendations (WMO-No. 1203, 2017). Consistency Verification (Double Mass Curve Method) — Following gap-filling, the annual consistency of each reconstructed series was verified by comparing cumulative annual totals against a reliable, gap-free reference station within the same HPR. The Double Mass method yielded coefficients of determination (R²) above 0.97 for all stations (maximum R² = 0.9998 for stations 444001–Coroatá and 543002–Parnarama; minimum R² = 0.9782 for station 646006–Grajaú), confirming the robustness of the reconstruction procedure and the spatial homogeneity of the dataset. The resulting series provides standardised monthly rainfall totals (millimetres) with full temporal coverage from January 1987 to December 2023 for all 100 stations, enabling direct intercomparison across the network and with gridded reanalysis and satellite-based precipitation products (e.g., CHIRPS, ERA5-Land, IMERG). Key Results from the Associated Publication Statistical analyses of the gap-filled and consistent series revealed the following main findings (Reschke et al., 2026): Spatial heterogeneity: Annual mean precipitation ranges from 1,036.4 mm (HPR9 — Chapadas das Mangabeiras) to 2,072.8 mm (HPR1 — Western Coastal Zone), reflecting a pronounced north–south gradient associated with the weakening of maritime influences and increasing continentality toward the interior. Trend analysis (Mann–Kendall test): No statistically significant monotonic trends were detected in any of the ten HPRs over the 1987–2023 period (p > 0.05 for all regions), consistent with the high interannual variability modulated by large-scale atmospheric teleconnections (ENSO, PDO, NAO). Sen's Slope Estimator: Nine of ten HPRs exhibited weak positive slopes (range: +1.46 to +9.98 mm year⁻¹), with the largest increase observed in HPR3 — Rosário and Itapecuru Mirim (+9.98 mm year⁻¹). HPR10 — Gerais de Balsas was the only region showing a negative slope (−0.99 mm year⁻¹), suggesting a potential reduction in water availability in the southernmost Cerrado-transition zone. Gap-filling performance: The Regional Weighting Method effectively reconstructed 314 missing monthly records across 74 stations, preserving statistical coherence, seasonality, and local rainfall patterns, with all R² values consistently above 0.97. Spatial and Temporal Coverage Attribute Details Domain Maranhão State, Brazil; border stations in Pará, Tocantins, and Piauí Coordinate system Geographic (WGS84 / SIRGAS 2000, EPSG:4674) Longitude range −48.21° to −42.23° W Latitude range −9.60° to −1.46° S Temporal coverage January 1987 – December 2023 Temporal resolution Monthly Number of stations 100 (90 ANA + 10 INMET) Homogeneous Precipitation Regions 10 (HPR1–HPR10) Total gap-filled records 314 monthly values across 74 stations Double Mass R² range 0.9782 – 0.9998 Homogeneous Precipitation Regions (HPRs) Region Designation No. of Stations Annual Mean Precipitation (mm) HPR1 Western Coastal Zone 13 2,072.8 HPR2 Baixada Maranhense Lowlands 8 1,766.3 HPR3 Rosário and Itapecuru Mirim 11 1,765.8 HPR4 Lower Parnaíba Maranhense 9 1,482.0 HPR5 Upper Mearim and Grajaú 12 1,334.8 HPR6 Caxias, Codó and Coelho Neto 5 1,463.5 HPR7 Imperatriz and Porto Franco 10 1,280.0 HPR8 Chapadas of Upper Itapecuru 15 1,191.5 HPR9 Chapadas das Mangabeiras 6 1,036.4 HPR10 Gerais de Balsas 11 1,375.9 Dataset Contents The dataset is distributed as a single Microsoft Excel workbook (MA-RAIN_v1.0.1_Organised.xlsx) structured in 28 worksheets: Sheet Content README Dataset description, metadata, and workbook navigation guide Station_Coordinates Coordinates, station codes, altitude, and data quality classification for all 100 stations INMET_Stations Coordinates of the 10 INMET meteorological stations ANA_Stations Coordinates of the 90 ANA rain gauge stations Monthly_Rainfall_All Complete raw monthly precipitation matrix — all stations × all months (including pre-1987 records where available) Monthly_Rainfall_Filled Gap-filled consistent monthly series per station (1987–2023) Climatological_Means Long-term (1987–2023) monthly and annual mean precipitation per station Annual_Totals_By_Region Annual precipitation totals per station, ready for trend analysis HPR1–HPR10 Station-level details and gap-filling residuals for each homogeneous precipitation region (10 sheets) MK_HPR1–MK_HPR10 Annual precipitation series per station within each region, with regional means, for Mann–Kendall trend analysis (10 sheets) Variables Variable Unit Description Monthly precipitation mm Total precipitation per calendar month, gap-filled and quality-controlled Annual precipitation mm Sum of monthly values per calendar year Monthly climatological mean mm Long-term (1987–2023) mean for each calendar month Annual climatological mean mm Long-term (1987–2023) mean annual total Regional mean annual precipitation mm Arithmetic mean of station annual totals per HPR Drought Severity Classification (SPI Reference) The dataset includes a Standardised Precipitation Index (SPI) drought severity classification for stations where the index was computed. Classification follows the widely adopted McKee et al. (1993) thresholds: Category SPI Range Near Normal −0.49 to +0.49 Mild Drought −0.99 to −0.50 Moderate Drought −1.49 to −1.00 Severe Drought −1.99 to −1.50 Extreme Drought ≤ −2.00 Intended Uses MA-RAIN v1.0.1 is designed to support: Climatological analysis — characterisation of the mean annual rainfall cycle, seasonal onset and cessation, and inter-annual variability across Maranhão; Trend detection — Mann–Kendall, Sen's Slope, and related non-parametric trend tests on monthly and annual series; Drought monitoring and assessment — computation of the Standardised Precipitation Index (SPI) and the Standardised Precipitation Evapotranspiration Index (SPEI); Hydrological and ecological modelling — provision of consistent boundary forcing for rainfall–runoff, water balance, and land surface models; Remote sensing validation — ground-truth and bias-correction of satellite precipitation estimates (CHIRPS, IMERG, ERA5-Land) over the Maranhão domain; Environmental change studies — detection of land-use and climate-driven shifts in the regional water cycle, including interactions with deforestation, fire, and secondary forest dynamics in the Brazilian Cerrado–Amazon transition zone; Agricultural planning and water resource management — support for drought risk assessment, irrigation planning, and adaptation strategies under climate change scenarios. Data Quality Statement All station series achieve full temporal coverage (January 1987 – December 2023) after gap-filling. The proportion of reconstructed values varies by station and is explicitly documented in the dataset. Stations with no gap-filled months are classified as Excellent Reference and serve as the primary anchor network for the regional regression models. Stations with few reconstructed values are classified as Excellent (n gaps), where n is the number of filled months. Users are advised to consult the quality flag associated with each station before applying the data to applications sensitive to reconstruction uncertainty. The consistency of all reconstructed series was verified through the Double Mass Curve method, yielding R² > 0.97 for all 100 stations, confirming that the gap-filling procedure preserved regional climatological coherence throughout the 1987–2023 period. ⚠ Version History and Change Log v1.0.1 — Current version (2026) This version supersedes v1.0 and introduces the following changes to the dataset file's structure and presentation. No precipitation values, station coordinates, station codes, or quality classifications have been altered. Workbook reorganisation — The original multi-sheet workbook was restructured into a standardised 28-sheet layout with consistent naming conventions, explicit section titles, and a dedicated README sheet providing full dataset documentation and navigation guidance. Full translation to English — All column headers, sheet names, variable labels, data-quality descriptors, drought severity categories, and month abbreviations have been translated from Portuguese to English. Station names, municipality names, and state designations are preserved in their original Portuguese form, as these are proper geographical names. Standardised column naming — Portuguese column headers (e.g., MUNICÍPIO (ESTAÇÃO), MESES COM OU SEM FALHAS DE DADOS, SEVERIDADE DA SECA (SPI)) were replaced with standardised English equivalents (Municipality (Station), Missing Data Assessment, Drought Severity (SPI)) across all sheets. Data-cell label translation — Portuguese categorical labels in data cells were systematically translated: quality descriptors (REFERÊNCIA EXCELENTE → Excellent Reference; EXCELENTE (1) → Excellent (1 gap)), SPI categories (Seca Fraca → Mild Drought; Seca Moderada → Moderate Drought; Seca Severa → Severe Drought; Seca Extrema → Extreme Drought; Quase Normal → Near Normal), and month abbreviations (JAN/FEV/MAR → Jan/Feb/Mar, etc.). Professional formatting — A consistent visual style was applied throughout: hierarchical colour-coded headers, alternating row striping, highlighted annual total rows, green-highlighted regional mean columns in Mann–Kendall sheets, and frozen header panes in all data sheets. v1.0 — Initial release (2025) Original release of the MA-RAIN dataset. Quality-controlled, homogenised, and gap-filled monthly precipitation series for 100 stations across Maranhão State, Brazil, covering January 1987 – December 2023. A total of 314 missing monthly records across 74 stations were reconstructed using the Regional Weighting Method and verified through the Double Mass Curve approach (R² > 0.97 for all stations). Reference Article When using this dataset, please cite both the dataset and the associated peer-reviewed publication: Dataset: Silva Junior, C.H.L., Reschke, G.A., Dias, C.W.S., Menezes, R.H.N. & Chagas, F.P. (2026). MA-RAIN v1.0.1: Maranhão Consistent Rainfall Dataset (1987–2023). Zenodo. https://doi.org/10.5281/10.5281/zenodo.20474136 Article: Reschke, G.A., Dias, C.W.S., Menezes, R.H.N., Chagas, F.P. & Silva-Junior, C.H.L. (2026). Filling the Gaps: Creating a Consistent Rainfall Dataset for Maranhão State, Brazil (1987–2023). Climate, 14(3), 63. https://doi.org/10.3390/cli14030063 Funding and Institutional Affiliation This dataset was developed in the context of the YbYrá-BR project — Space-Time Quantification of CO₂ Emissions and Removals by Brazilian Forests — and associated research activities. Formal financial support was provided by: CNPq — National Council for Scientific and Technological Development, Brazil (Processes 304664/2024-3, 400634/2024-4, and 401741/2023-0; C.H.L. Silva-Junior); CAPES — Coordination for the Improvement of Higher Education Personnel, Finance Code 001. Acknowledgements: The authors thank the State University of Maranhão (UEMA) and the Federal University of Maranhão (UFMA) for institutional support, and the National Water and Sanitation Agency (ANA) and the National Institute of Meteorology (INMET) for providing the rainfall datasets. Scholarship support from the Maranhão Research Foundation (FAPEMA) and CAPES is also gratefully acknowledged. Institutional affiliations: Instituto de Pesquisa Ambiental da Amazônia (IPAM) — Núcleo de Carbono / CCAL Universidade Federal do Maranhão (UFMA) — BIONORTE Network Graduate Program / Graduate Program in Biodiversity and Conservation (PPGBC) Universidade Estadual do Maranhão (UEMA) — Graduate Program in Agricultural Sciences / Department of Agricultural Engineering Keywords monthly precipitation · rainfall dataset · Maranhão · Brazil · gap-filling · Regional Weighting Method · Double Mass Curve · homogeneous precipitation regions · Mann–Kendall test · Sen's slope · climate variability · drought · SPI · ANA · INMET · tropical climatology · water resources · Cerrado · Amazon · climate change Licence This dataset is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence. You are free to share and adapt the material for any purpose, provided appropriate credit is given, a link to the licence is provided, and any changes made are indicated. Full licence text: https://creativecommons.org/licenses/by/4.0/