How does the reference period influence meteorological drought analysis and monitoring? A case study in Northeast...

Droughts are extreme climatic events that compromise water, food, and energy security worldwide. Monitoring such events is essential for understanding and mitigating their impacts. However, data scarcity and the lack of consensus on the appropriate reference period for drought analysis remain persistent challenges. This study examines how different reference periods influence the identification and characterization of recent meteorological droughts (2010-2019) in Northeast Brazil (NEB) using the Standardized Precipitation Index (SPI). A total of 602 precipitation series from the Climate Research Unit (CRU), uniformly distributed across the region over a century (1920-2019), were analyzed. Drought metrics were evaluated at multiple temporal (monthly, quarterly, semiannual, and annual) and spatial (grid and state) scales, comparing nine alternative temporal scenarios (10-100 years) with a conventional 30-year reference period. During 2010-2019, several states experienced drought in more than 80 % of events, with over 10 % classified as extremely dry under SPI-12. Average drought duration exceeded 36 months in multiple states, and mean intensity approached 1.5 per month. Scenarios S2 (20 years) and S4 (40 years) best matched the 30-year baseline, with R, Kd, and Kp values above 0.80 and RMSE below 0.20. Longer scenarios overestimated drought duration by >10 months, severity by >10 SPI units, and dry-event frequency by >10 %. The shortest scenario, S1 (10 years), showed the poorest performance, with RMSE near 1.0, Kp below 0.30, and marked underestimation of duration, severity, and frequency. These results highlight the critical role of selecting an appropriate reference period and provide a statistically robust and adaptable methodology to improve drought monitoring and early warning systems, particularly in data-scarce regions, thereby supporting the design of more effective public policies for managing extreme events. This integrative multi-metric framework fills a critical gap in understanding reference-period sensitivity.