Evaluations of Precipitation, Temperature, and Circulation Characteristics in Simulations of CMIP6 Models over the Arabian Peninsula

Abstract

This study evaluates the performance of 74 CMIP6 (including HighResMIP) models for the historical climate of the Arabian Peninsula (AP). The analysis examines whether models reproduce the mean climate, trends (accounting internal-variability), extremes, and key circulation/process diagnostics. Despite strengths, several limitations persist across key aspects. Precipitation biases include annual-mean wet-bias over the southern AP and dry-bias over the central AP, and models underestimate the frequency and intensity of heavy-precipitation-events. Temperature simulations show high-pattern agreement, but nearly all models exhibit wet-season cold-bias and misrepresent the hot- and cold-tails. Distribution diagnostics indicate that inter-model-spread dominates uncertainty and that ensemble-means can be sensitive to outliers in extreme analyses. Precipitation biases covary with pressure-gradients, Red Sea flow/convergence, synoptic-transients, and jet-structure; while dry-season biases are dynamically consistent with a monsoon-desert linkage diagnosed using ω500, the tropical-easterly jet, and Tokar-gap related flow. Coarse-topographic representation in models weakens winds and moisture transport, but its impact is also subject to season, variable, and orographic height. Paired within-family runs further show that higher-resolution and added Earth-system-complexity can improve AP fidelity in some configurations but can also yield neutral or degraded performance, indicating that resolution is beneficial but not sufficient for robust skill. No single model excels across all diagnostics, but a combined phenomenology-process-framework identifies high-skill models while flagging cases where strong composite performance may mask weak process realism or bias compensation. These results provide an observationally-constrained basis for model selection and uncertainty characterization, underscoring the need for rigorous validation and continued model improvement for reliable regional-climate projections.