Abstract: As a critical component of the global water cycle, atmospheric water vapor trends and driving mechanisms are of significant importance to agricultural production and ecosystem stability. Based on multi-source data from 1959 to 2023, this study integrates continuous wavelet transform, cross-wavelet transform, multivariate covariance testing, and interannual variability quantification methods to systematically analyze the primary drivers of global atmospheric water vapor, including surface variables and oceanic dynamic factors, while elucidating their regional impacts on major agricultural production zones. Results indicate a long-term increase in water vapor at 0.0168 kg/(m²·yr) (1959–2023), with intense fluctuations in the tropics (up to 26 kg/(m²·yr) in summer), stable trends at mid-to-high latitudes (23–25 kg/(m²·yr)), and low levels in polar regions (< 23 kg/(m²·yr)). Short-term cycles (6–16 mon) show strong seasonal synchronization with temperature and precipitation, while long-term trends (104–256 mon) are dominated by climate change and oceanic influences. For enhanced forecasting, we propose a hybrid model integrating discrete wavelet transform with gated recurrent units. This framework outperforms benchmark models (MAE = 0.185, R² = 0.957), improving R² by 27.5% compared to the standalone model. Predictions for the next 24 mon indicate an overall slow increase in water vapor, peaking at 27 kg/m² in summer. This approach markedly improves prediction reliability, providing robust support for optimizing agricultural irrigation across regions and mitigating meteorological disasters.