Probabilistic Changes of Compound Dry-to-Wet Events: Detection and Uncertainty from Climate Model Ensembles

Compound hydroclimatic extremes, particularly dry-to-wet transitions, represent a growing climate risk due to their cascading impacts on flooding, agriculture, and water resources. Under climate change, shifts in the frequency and severity of such compound events are expected, yet large uncertainties remain in their detection, characterization, and future evolution. This study presents a probabilistic, ensemble-based assessment of compound dry-to-wet events across Pakistan, with explicit attention to event detection, severity classification, and uncertainty in climate projections. Compound dry-to-wet events are detected using the Standardized Precipitation Evapotranspiration Index (SPEI), capturing transitions from sustained dry conditions to subsequent wet extremes. To systematically characterize event severity, we develop a compound magnitude index that integrates the severity and duration of both the dry and wet phases of each event. This index enables the classification of compound dry-to-wet events into mild, moderate, severe, and extreme categories, facilitating robust comparisons across regions, models, and emission scenarios. The analysis is based on historical and future simulations from CMIP6 global climate models and CORDEX dynamically downscaled regional climate models under multiple Shared Socioeconomic Pathways (SSPs). Changes in the frequency, duration, intensity, and severity distribution of compound dry-to-wet events are evaluated relative to a historical reference period. Probabilistic metrics are used to quantify ensemble agreement and spread, while uncertainty is decomposed into contributions from model structure, scenario choice, and internal climate variability. Differences between CMIP6 and CORDEX ensembles are further examined to assess the role of regional downscaling in representing compound event characteristics. Results indicate an increased likelihood and severity of compound dry-to-wet events under higher-emission scenarios, with pronounced spatial heterogeneity across Pakistan. In particular, severe and extreme events show more robust increases than mild and moderate events. Model uncertainty dominates projections of compound event magnitude, while scenario uncertainty becomes increasingly important toward the late 21st century. Regional climate models enhance the representation of localized extremes but exhibit larger inter-model variability. This study advances compound event research by introducing a SPEI-based compound magnitude framework and a comprehensive uncertainty assessment, providing valuable insights for climate risk assessment and adaptation planning in climate-vulnerable regions.