Sustainable management of surface water resources requires continuous monitoring and accurate analysis of water quality at various time intervals. This study assessed the water quality of the Talar River during low-water (September) and high-water (May) periods. During field visits, 15 sampling points were identified in areas with different land uses. Samples were collected from depths ranging from 15 to 30 cm below the water surface on two occasions. The results showed that, during the high-water period, electrical conductivity (EC) and total dissolved solids (TDS) exhibited a strong positive correlation with dissolved ions, such as calcium (Ca²⁺), magnesium (Mg²⁺), and sodium (Na⁺). Additionally, pH exhibited a negative correlation with heavy metals, such as aluminum (r = -0.35) and silicon (r = -0.28), indicating changes in the solubility of these elements under acidic conditions. Principal component analysis (PCA) revealed that, during the dry period, two main components accounted for over 90% of the variance in the data. The first component was influenced by dissolved ions such as sodium, chloride, and sulfur. The second component was related to parameters such as EC and turbidity. During the high water period, the first component was primarily influenced by TDS and EC, accounting for 95.69% of the variance. Modeling calcium concentration using deep learning models revealed that the convolutional neural network (CNN) model outperformed the long short-term memory (LSTM) model. At the Shirgah station, the CNN model fit the real data better, with a correlation coefficient of 0.75; the LSTM showed a coefficient of 0.6.