Abstract: Multiple challenges exsit in the study of rice water requirement (RWR), such as the handling of massive data and the complexity of spatio-temporal scale changes. These challenges make it difficult to capture key characteristics of RWR using a single approach. Therefore, to overcome these challenges, we proposed an innovative approach combining time-domain and frequency-domain for identifying the key influencing factors of water requirements in this study. The Penman-Monteith formula, based on seven meteorological indices and four circulation indices, was used to analyze the characteristics of RWR including crop water requirement (ET_c) and irrigation water requirement (IR) during the rice growth period in the Gaoyou irrigation district from 1980 to 2021. From the perspective of time-frequency domain, this study integrated Pearson correlation analysis, wavelet analysis, projection length analysis, and M-K test methods, in combination with energy partitioning analysis, to propose a method for analyzing key influencing factors of RWR and identifying their variation trends. The results of the study showed as follows. 1) In terms of interannual variation, the multiyear mean of ET_c and IR were 532.88 mm/a and 285.04 mm/a, respectively. Their variations ranged from 444.61 to 638.55 mm/a for ET_c and 32.63 to 481.62 mm/a for IR. IR showed greater interannual variability than ET_c. The interannual anomaly values analysis showed that ET_c and IR transitioned from below average to above average around the year of 2000. And, the monthly variation during the rice growth period showed that RWR was the highest in August and the lowest in October each year. 2) Both ET_c and IR had two main energy zones, namely Zone I and Zone II. Zone I had a larger time scale and longer periods compared to Zone II. For ET_c, Zone I had cycles of 25-30 years, while short-period components in Zone II had cycles of 5-15 years. And for IR, the long-period signal components had cycles of 13-18 years, and the short-period components had cycles of 3-9 years. In Zones I and II, ET_c was primarily influenced by relative humidity and sunshine duration, with relative humidity leading ET_c by approximately half a cycle and sunshine duration showing no phase difference with it. IR was primarily influenced by precipitation in both Zones I and II, with a phase difference of half a cycle in both cases. 3) The analysis of energy partitioning and correlation results indicated that in Zones I and II, ET_c was influenced by relative humidity and sunshine duration. Relative humidity led ET_c by approximately half a cycle, while sunshine duration had no phase difference with ET_c. IR was primarily influenced by precipitation in both zones, with a phase difference of half a cycle. ET_c showed highly significant negative correlations with sunshine duration and positive correlations with relative humidity. IR was negatively correlated with precipitation. Overall, sunshine duration and relative humidity were identified as key influencing factors for ET_c, while precipitation was the key factor for IR. Both ET_c and IR showed a "main shock orderly, aftershocks continuous" pattern in relation to their key influencing factors. 4) Wavelet analysis revealed that ET_c was currently in an upward trend after a period of low oscillation, while IR showed an upward trend from a trough. The M-K test results aligned with these findings, though they indicated that the upward trend was slow and accompanied by periodic fluctuations. The M-K mutation test identified abrupt changes around the year of 1990 for both ET_c and IR. After that, both ET_c and IR exhibited an increasing trend, which became more pronounced after the year of 2005. The research results provide a reference for the rice irrigation system in the Gaoyou irrigation district and enhance the reliability of identifying key influencing factors of RWR, offering significant application prospects in other regions or related fields.