Extreme Precipitation Indices Trends and Patterns in the Philippines

Arnold Salvacion , Allen Nazareno, Destiny Lutero

The Philippines experiences approximately 86% of its recorded disaster events from extreme precipitation, yet national-scale trend studies remain constrained by sparse station networks and narrow index coverage. This study mapped the spatial patterns and long-term trends of 12 extreme precipitation indices across the Philippine archipelago using 45 years of daily gridded precipitation data from the Climate Hazards Group InfraRed Precipitation with Stations dataset over the period 1981 to 2025. Trend magnitude and direction were estimated at each grid cell using the Theil-Sen slope estimator and assessed for statistical significance using the Mann-Kendall test; spatial clustering of trends was examined using the Local Indicator of Spatial Association. Field-significance testing (Monte Carlo permutation) and a Hamed-Rao (1998) autocorrelation correction were subsequently applied to determine which spatial trend patterns are distinguishable from spatially correlated noise. Significant increasing trends in total annual precipitation covered 71.9% of land grid cells (Hamed-Rao-corrected; 60.5% raw) with a median Sen's slope of 17.6 mm yr⁻¹, concentrated predominantly in Mindanao and the Visayas, with central Luzon and southern Luzon (Bicol) as secondary contributors, and this pattern was field-significant (p = 0.004). Heavy precipitation frequency showed the most spatially extensive and most robust signal, with significant increasing trends in heavy rainfall days covering up to 73.1% of land grid cells (R10; field-significant, p = 0.010). Re-analysis found mean daily rainfall intensity on wet days (SDII) to be predominantly increasing, not decreasing as initially reported, across 25.6% of land grid cells, though this pattern did not reach field significance (p = 0.058) and is therefore suggestive rather than conclusive. Rainfall days are consequently becoming more frequent across much of the country with high statistical confidence, while evidence for a corresponding change in per-event intensity remains inconclusive, a more conservative conclusion than a straightforward frequency-versus-intensity dichotomy. The contrasting trend directions between 3-day and 5-day maximum accumulations are suggestive of, but do not on their own demonstrate, a lengthening of sustained rainfall episodes, since the 5-day signal did not reach field significance (p = 0.115). Spatial clustering analysis, computed on the complete spatial field of Sen's slopes, identified hotspot clusters concentrated on the typhoon-exposed eastern seaboard, including western Mindanao and the Zamboanga Peninsula for the two most robust frequency indices (R10, R20), and a coldspot cluster concentrated in northern Luzon; these results inform flood risk management and agricultural water planning across the exposure-sensitive sub-regions identified.