Assessing Probable Maximum Precipitation in the Soummam Basin (Algeria) Using Bias‐Corrected Climate Projections

Probable Maximum Precipitation (PMP) represents a major hydro‐meteorological hazard in Algeria, with its frequency and intensity increasing under the effects of climate change, thereby heightening threats to populations, infrastructure, and ecosystems. Yet existing PMP projections rarely employ multi‐variable bias correction or systematically compare correction methods, which is a critical gap given PMP's dependence on multiple correlated variables. The Soummam watershed, located in northeastern Algeria, is particularly vulnerable to extreme hydro‐meteorological events. In this context, this study analyzes the future evolution of PMP by correcting biases across four variables used for PMP estimation (precipitation, minimum temperature, relative humidity, and wind speed) using three distinct methods: Quantile Mapping (QM), Quantile Delta Mapping (QDM), and CDF‐t transformation. Using daily observations from 56 rain gauge stations and ERA5‐Land data, we drive bias‐corrected CMIP6 climate models for two periods (2026–2064 and 2065–2100) under SSP2‐4.5 and SSP5‐8.5 scenarios. Results reveal substantial method‐dependent variations in PMP projections. The QDM and CDF‐t methods project PMP increases of +65% to +90%, with large ensemble uncertainty (CV: 60%–90%, statistically significant in 70% of comparisons, ANOVA p < 0.05), predominantly in the 2026–2064 period under SSP2‐4.5; by contrast, QM produces decreases of up to −40%, suggesting fundamental differences in how methods treat future distributional changes in climatic variables critical to PMP calculations. The counterintuitive greater SSP2‐4.5 increases reflect the complex interplay between those variables too. KDE distributions exhibit bimodality with QDM and CDF‐t, reflecting heterogeneous model and elevation‐dependent responses across the CMIP6 ensemble, whereas QM produces narrower, unimodal distributions. Spatially, western and northwestern high‐altitude sectors show the largest increases, while other regions remain stable or decline. These findings underscore how univariate bias correction methodology fundamentally shapes PMP projections and highlight the necessity of ensemble‐based uncertainty framing in climate‐driven hydrological risk assessment.