The operation of cascade reservoirs has a significant impact on river water temperature. This study constructs a quantitative attribution framework to distinguish the impacts of climate change and reservoir operation on river water temperature by integrating the Maximal Overlap Discrete Wavelet Transform (MODWT), Multiresolution Analysis (MRA), ERA5-Land reanalysis data, and a Bi-LSTM model. Taking the Xiangjiaba Station in the lower Jinsha River as a case study, the multi-scale effects of cascade reservoir operation on water temperature were investigated. The results indicate that: (1) The joint operation of the cascade reservoirs significantly dampens water temperature fluctuations at medium-to-high frequencies (daily, weekly, monthly scales), exhibiting a "smoothing effect"; (2) At low frequencies (seasonal and inter-annual scales), reservoir operation leads to a significant "attenuation effect" and "lag effect" on water temperature, and these effects intensify with an increasing number of reservoirs in operation; (3) Reservoir operation is the dominant factor driving water temperature changes at the seasonal scale, while climate change primarily drives the changes at the annual scale during the four-reservoir joint operation period. This study reveals the multi-temporal scale characteristics of the impact of cascade reservoirs on water temperature, proposes a novel attribution analysis method, and provides a scientific basis for the ecological operation of reservoirs and watershed water temperature management.