نبذة مختصرة : Upper 99th percentile hourly, 90th percentile daily, and 90th percentile pentad rainfall values were calculated over the four largest Japanese islands for June–July every year in the 1952–2010 period, using 10 ensemble members of the 5 km resolution d4PDF (database for Policy Decision-making for Future climate changes) climate ensemble and 126 rain gauges. The HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise) algorithm was used to cluster rain gauges and d4PDF grid points, based on a multi-frequency metric. Six analysis regions were identified based on rain-gauge clusters. Spearman correlation was calculated between cluster rainfall extremes and standardised scores of five modes from the varimax-rotated extended principal component analysis of Pacific sea surface temperature (SST) anomalies. By order of explained variance for the analysis period, these modes represent El Niño–Southern Oscillation (ENSO) growth, ENSO decay, warming trend mixed with climate variability (Trend+), non-canonical ENSO, and other Pacific decadal variability (PDV). Rain gauges showed a field-significant correlation between hourly extremes and Trend+ and an anti-correlation between daily extremes and PDV; d4PDF showed excessive spatially widespread correlation and anti-correlation with ENSO decay and non-canonical ENSO, respectively. Rainfall extremes were better related to indices describing the regional monsoon jet's position and water vapour flux for both rain gauges and d4PDF. Modulation of the regional monsoon by the Pacific SST modes was used to explain the relationship between rainfall extremes and Pacific SST modes. The differences between rain gauges and d4PDF may be caused by slower monsoon development in d4PDF.
Processing Request
No Comments.