Accurate reference evapotranspiration (ET₀) estimation is critical for sustainable irrigation management in tropical monsoon climates, yet systematic comparisons distinguishing correctable bias from structural inadequacy remain limited. This study evaluated Modified Penman (MP), Blaney-Criddle (BC), and FAO Penman-Monteith (PM) methods using 15-year (2010-2024) monthly climate data in Subak Balangan, Bali, employing multi-criteria validation with Nash-Sutcliffe Efficiency (NSE) decomposition. Penman-Monteith demonstrated superior performance with excellent correlation (r=0.852), lowest errors (MAE=0.719 mm/day), and best relative accuracy (MARE=15.2%). Modified Penman revealed a critical paradox: worst NSE (-8.914) yet highest correlation (r=0.870), with decomposition showing 86% of error from correctable systematic bias. Calibration (factor 0.78) improved NSE to 0.684, demonstrating viability for data-limited contexts. Blaney-Criddle proved unsuitable with a negative correlation (r=-0.782) and phase misalignment from the temperature-only formulation. PM represents the optimal choice with complete data; calibrated MP offers a practical alternative for resource-limited tropical irrigation systems. This diagnostic framework provides actionable guidance for irrigation managers to optimize water allocation and cropping schedules, contributing to food security and sustainable water management in traditional agricultural systems throughout tropical regions.