Precipitation variability at different temporal scales in western Pará: analysis of observed and estimated data

Precipitation exhibits strong spatial and temporal variability across the Amazon, where sparse meteorological networks and short homogeneous records constrain characterization of rainfall regimes and hydroclimatic variability. We evaluated MERGE and the Brazilian Daily Weather Gridded Data (BR-DWGD) against observations from the Brazilian National Institute of Meteorology at Altamira, Belterra, Itaituba, Monte Alegre, and Porto de Moz. The common period from 1998 to 2017 was assessed at monthly, quarterly, semiannual, and annual scales using error, correlation, efficiency, and agreement metrics. The Standardized Precipitation Index was evaluated at SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12, and hierarchical clustering assessed similarities among precipitation regimes. The 1961–2025 BR-DWGD series was analyzed through Continuous Wavelet Transform and compared with the Oceanic Niño Index to examine long-term variability and correspondence with the El Niño-Southern Oscillation. MERGE showed the strongest mean monthly performance, with Pearson correlation of 0.970, Nash-Sutcliffe efficiency of 0.936, and Willmott agreement index of 0.983, compared with 0.910, 0.822, and 0.950 for BR-DWGD. Performance improved through the semiannual scale but declined annually. MERGE also showed greater SPI agreement and classification accuracy. The annual cycle dominated long-term precipitation variability, with intermittent interannual structures of approximately four to eight years. Negative anomalies occurred more frequently during El Niño and positive anomalies during La Niña. MERGE better represented observed precipitation, whereas BR-DWGD provided the temporal continuity required for long-term analysis.