This study aimed to use artificial intelligence to forecast variables related to the movement of water out of predefined areas using knowledge of the variables related to water moving into these same areas. The data for this study was hourly numerical data processed from the National Water Model database. The artificial intelligence models used for these predictions were then analyzed to gain a more comprehensive understanding of their inner workings, how the predictions were made, and how they could be improved. This process also presented the opportunity to explore how hydrological information can be encoded in and eventually extracted from artificial intelligence models. The results showed the models were successful in making predictions, and further analysis showed how different geological and hydrological properties influenced the models' performance.