Abstract: Groundwater level driving factors are the most important components of the national water resources in modern agriculture. Sustainable water use can be expected to regulate in the arid and semi-arid regions. In this study, a systematic investigation was implemented on the dynamic characteristics of the groundwater level. Their controlling factors were also quantified after optimization. The groundwater serves as a critical water supply source. The study area was designated as the plain area of Jimsar County, Changji Hui Autonomous Prefecture, Xinjiang, China. A multi-source dataset was integrated, including the long-term groundwater level monitoring from 2003 to 2023, meteorological data (e.g., precipitation and temperature), surface runoff, land use/land cover change, and human activity intensity evaluation. Specifically, the Kriging interpolation was utilized to map the spatial distribution of the groundwater levels. While the trend analysis was applied to capture the temporal evolution patterns, thereby characterizing the spatiotemporal variations of the groundwater level, Wavelet coherence analysis was employed to explore the time-frequency correlation between groundwater level and hydrometeorological factors. Bivariate spatial autocorrelation was used to examine the spatial coupling relationship between human activities and groundwater level variations. The human activity intensity index was adopted to assess the degree of human disturbance. There were the driving mechanisms of the groundwater level. Additionally, a geographical detector model was applied to quantify the individual and interactive contributions of each influencing factor to groundwater level dynamics. The results indicated that the intra-annual dynamic variation of the groundwater level was predominantly dominated by an exploitation-driven pattern, which was closely associated with the seasonal characteristics of the agricultural irrigation. Meanwhile, the long-term groundwater level shared an overall fluctuating downward trend. Four stages were divided: A relatively stable fluctuating variation (2003–2008), a rapid and significant decline (2009–2014), a slight recovery and upward trend (2015–2017), and a reversion to a gradual downward trend (2018–2023). Significant resonance periods and time-lag effects were observed between groundwater level and precipitation/runoff at different time scales, indicating a significant positive correlation. The groundwater level was lagged behind precipitation and runoff by approximately 0–2 months at the phase angle from 0° to 60°. The gradual expansion of the cultivated land area was driven by agricultural development from 2003 to 2023. The regions with high-intensity human activities shared a continuous expansion trend. A local spatial autocorrelation existed between human activity intensity and groundwater level, indicating a strong spatial coupling between human disturbance and groundwater evolution. Additionally, the dynamic groundwater level was mainly influenced by multiple factors, including surface runoff, human activities, precipitation, and land use types. Notably, the interactions of these factors exerted a more significant synergistic impact on the groundwater level dynamics, compared with their individual effects. The findings can greatly contribute to the dynamic response mechanisms of the groundwater in the arid areas. Theoretical support and practical technical references can also be used to formulate the regional groundwater indicators, and then realize the long-term sustainable utilization of the groundwater resources.