This study focuses on methods for the prediction of rainfall to support sustainable management of water resources in Aizawl, since rainfall patterns do significantly influence water availability. Therefore, a comparative assessment was done with a range of Machine Learning Algorithms: Random Forest, Gated Recurrent Unit, Extreme Gradient Boosting (XGBoost) and Long Short-Term-Memory. The results show that XGBoost is more commanding as compared to the other models with 2.94 as root mean square error value and 0.87 as mean absolute error value. An R2 value of 0.95 shows that the model accounts for 95% of the variability in the rainfall data, meaning it has very strong predictive power. Additionally, an Nash-Sutcliff efficiency of 0.95 with a more dependable prediction performance. The implication of the above results is that the ensemble model presents a promising method to forecast rainfall patterns in an accurate way, which is important for sustainable management of water resources and preparing for climate variability in Aizawl. These insights could feed into policy recommendations and adaptive strategies aimed at improving water security and resilience in Aizawl.