<![CDATA[ Rainfall plays a crucial role in shaping the weather in Indonesia, influenced by factors such as latitude, elevation, wind patterns, land and water distribution, as well as topography. Rising temperatures contribute to the increased intensity of extreme rainfall, amplifying the potential risk of disasters. Therefore, it is necessary to conduct analyses to predict weather based on historical time series data. This study aims to identify short-term rainfall patterns and trends using the SARIMAX method. The initial stage involves data processing and splitting the data into training data (2005-2014) and test data (2015-2019). Time series decomposition is then performed to identify patterns, followed by period identification and stationarity testing using the ADF test. The SARIMAX model is selected based on the lowest AIC value, followed by parameter estimation and diagnostic tests. Rainfall predictions are evaluated using model performance evaluation methods, while inferential statistics are used to describe population attributes through confidence intervals. The forecasting results show that using the SARIMAX (0, 0, 1)(0, 1, 1, 12)12 model for short-term rainfall prediction achieved the best performance with the smallest MSE, MAE, and MAPE values on the test sample with a 95% confidence level. Repeating rainfall patterns with a mid-year decline and significant variation in magnitude were identified from the sample data for August 2020-2027, with the lowest rainfall at 39.05 mm and the highest at 397.03 mm in December. The implications of this study support efforts to mitigate natural disasters due to unexpected weather changes by integrating this model into early warning systems and water resource planning.Keywords : hydrology, precipitation, time series, short-term, SARIMAX ]]>
