Toward Reducing Uncertainty in Simulating Temporal Clustering of Extreme Precipitation in Morocco: Insights from High-Resolution GCMs

Abstract Temporal clustering of extreme precipitation (TCEP)—multiple extreme precipitation events (EPEs) occurring within weeks—poses compound flood risk in Morocco, yet its climatology, drivers, and representation in global climate models (GCMs) remain poorly characterized. We develop a process-based framework using Multisource Weighted-Ensemble Precipitation (MSWEP) (0.1°, 1979–2025) and ERA5 reanalysis to detect and diagnose TCEP and to evaluate six CMIP6 High-Resolution Model Intercomparison Project (HighResMIP) models. Specifically, we detect TCEP episodes with a nonparametric, count-based approach that identifies clusters of EPEs, daily exceedances of the 99th-percentile threshold, occurring within a 21-day moving window; statistical significance of this clustering is assessed via bootstrap resampling. TCEP drivers are diagnosed from composites of atmospheric anomalies filtered with the false discovery rate (FDR). Our findings show that TCEP is statistically significant in northwestern and southeastern Morocco, with episodes peaking from November to April and reaching up to 283 mm per event (≈47%–99% of the mean annual total) and up to four EPEs per episode. In the northwest, TCEP is associated with enhanced westerly integrated vapor transport (IVT), frequent atmospheric river (AR) landfalls, a quasi-stationary trough over the eastern North Atlantic–Iberia, and an equatorward-displaced jet, often during negative North Atlantic Oscillation (NAO) phases; in the southeast, anomalies are weaker and orographic forcing plays a larger role. HighResMIP models reproduce EPE counts per episode but underestimate total episode rainfall by up to 52%. A clustering simulation skill index (CSSI) ranks FGOALS-f3-H, MPI-ESM1-2-XR, and HiRAM-SIT-HR highest, and BCC-CSM2-HR ranks lowest across all regions. Comparisons between coupled ocean–atmosphere and AMIP-style (prescribed SSTs) simulations indicate that ocean coupling is an important source of TCEP bias.