Urban green spaces function as essential thermal mitigation zones in arid regions where extreme heat events increasingly threaten public health and urban livability. This study integrates field monitoring campaigns with computational fluid dynamics modeling to examine the coupled effects of vegetation arrangement, water feature geometry, and terrain morphology on microclimatic conditions across three climatically distinct arid cities in China: Turpan (Xinjiang), Xi'an (Shaanxi), and Hohhot (Inner Mongolia). Multi-season field observations (summer and transitional periods, 2023-2024) captured air temperature, humidity, wind velocity, and solar radiation at 15-point monitoring arrays within representative urban parks. Numerical simulations employed ENVI-met v5.6 coupled with SkyHelios 1.2 for enhanced radiative flux calculations, validated against empirical datasets (R-2 = 0.89-0.94 for temperature; R-2 = 0.82-0.87 for wind speed). Findings demonstrate that strategic vegetation placement reduced mean radiant temperature by 18-24 degrees C and lowered Physiological Equivalent Temperature (PET) by 5.2-7.8 degrees C during afternoon peak hours across the three cities, with deciduous species outperforming evergreens by 1.8-2.4 degrees C in summer cooling efficiency. Water features exhibited non-linear cooling responses: surface areas representing 15-25% of park footprint generated optimal daytime PET reductions (3.9-5.6 degrees C) while minimizing nocturnal heat retention penalties (<1.2 degrees C PET increase). Comparative thermal mapping revealed that distributed small-scale water elements (5-8 features of 200-400 m2 each) surpassed single large water bodies by 12-18% in spatial cooling effectiveness. Topographic analysis identified that north-facing slopes with 3-7 degrees gradients combined with strategic tree positioning achieved 23-31% greater cooling duration compared to flat terrain, extending thermal comfort periods by 2.5-3.7 h daily. Multi-objective optimization incorporating genetic algorithms identified Pareto-optimal landscape configurations: 68-76% canopy coverage with heterogeneous density patterns, 18-24% water surface area in clustered distributions, and integrated micro-topography with 4-6 degrees undulations. These configurations delivered 35-42% improvement in diurnal thermal comfort metrics (percentage of hours with PET <35 degrees C) compared to conventional uniform designs. Sensitivity analysis revealed vegetation spatial arrangement contributed 47-53% of total cooling variance, followed by water feature distribution (28-34%) and topographic modification (15-21%). Cross-city validation demonstrated that these optimization principles maintain effectiveness across varying background climates (temperature range adaptability: 32-44 degrees C daily maximum). This research advances evidence-based frameworks for climate-responsive urban park design in water-scarce regions, providing quantitative guidelines for landscape architects and urban planners to enhance outdoor thermal comfort while addressing site-specific constraints and multi-functional requirements.