Abstract
<title>Abstract</title> <p>Tropical archipelagos are among the most heterogeneous rainfall environments on Earth, where complex island topography, surrounding warm seas, and multi-scale atmospheric circulation interact to produce sharp spatial gradients in precipitation over very short distances. In the Maluku Archipelago, the scarcity of long-term, spatially continuous ground-based observations has limited systematic characterisation of daily rainfall persistence and dry–wet spell dynamics, even though such information is central to hydrological risk management in small, water-limited island systems. This study quantifies the spatiotemporal variability of daily rainfall persistence across the Maluku Archipelago using ERA5 reanalysis data and a first-order, two-state Markov chain framework for the period 2000–2025. Daily precipitation at eight representative climate zones, spanning mountainous interiors, coastal plains, and open-ocean and southern island settings, was classified into wet and dry states using a 1 mm day⁻¹ threshold, from which monthly transition probabilities, stationary probabilities, expected spell durations, and the 90th and 95th percentile (P90, P95) extreme-rainfall thresholds were derived. The results reveal pronounced spatial heterogeneity in rainfall persistence across the archipelago: the mountainous interior of Seram Island recorded both the highest mean daily precipitation (11.85 mm day⁻¹) and the highest extreme threshold (P95 = 72.8 mm), consistent with strong orographic enhancement, whereas the southern Tanimbar region (Saumlaki) recorded the lowest mean daily precipitation (4.45 mm day⁻¹), consistent with the seasonal intrusion of dry continental air from northern Australia during the Southeast Monsoon. Monthly transition matrices captured a systematic seasonal evolution of dry-to-dry and wet-to-wet persistence, with expected wet-spell durations peaking in June (mean 8.87 days; P90 = 21.3 days) and dry-spell durations peaking in July (mean 6.72 days; P90 = 14.8 days), and these patterns differed markedly among the eight climate zones. Together, these findings demonstrate that rainfall persistence over the archipelago is jointly governed by monsoonal circulation and island-scale physiographic complexity rather than by either factor alone. By coupling high-resolution reanalysis with a parsimonious stochastic framework, this study provides a transferable, reproducible approach for quantifying atmospheric memory in data-sparse tropical island regions, with direct relevance to drought monitoring, flash-flood early warning, and climate adaptation planning across the Indonesian Maritime Continent.</p>