Glacial lake outburst floods (GLOFs) have become increasingly frequent under climate warming. Yet the links between lake surface temperature (LST) dynamics and GLOF triggers remain poorly understood due to the absence of in situ lake temperature observations. This study investigates the potential of MODIS-derived LST to serve as a precursor for GLOFs at Lake Merzbacher, a frequently outbursting ice-dammed lake. We analyzed LST trends from 2000 to 2022 and examined their short-term dynamics preceding 25 documented GLOF events. Our results reveal a significant summer LST warming trend of 0.06 °C·yr⁻¹, exceeding the regional air temperature rise. We identified a critical LST threshold of 12 °C, with ~90 % of GLOFs occurring above this level. More importantly, we detected distinct thermal precursors: a rapid LST increase (peaking at 0.65 °C·day⁻¹) beginning ~8 days before outburst, and a critical acceleration phase (exceeding a threshold of 1.04 °C·day⁻²) around 9 days pre-GLOF. Furthermore, the peak discharge of floods showed the strongest correlation with the 15-day cumulative LST before outburst (r = 0.77), highlighting the role of integrated thermal energy in controlling flood magnitude. This study establishes LST not merely as a background climate indicator but as a source of diagnostic, short-term warning signals. We propose a multi-parameter framework integrating absolute LST, its rate of change, and acceleration to enhance early-warning systems for ice-dammed lakes under climate warming.