Operational streamflow forecasting is crucial for managing water resources in Arizona. A new system was developed using machine learning and meteorological forecasts to predict streamflow for up to 35 days. By integrating short-term streamflow forecasts with seasonal water supply forecasts, the accuracy of predictions improved, helping to balance water conservation and flood risk. The study showed that incorporating 35-day streamflow predictions led to better seasonal forecasts, especially in early winter, compared to using 7-day predictions. This highlights the value of using subseasonal to seasonal forecasts for improving long-term water resource planning in semiarid watersheds.