Snow cover is a significant component of the hydrological cycle affecting stream discharge through snowmelt and soil moisture. Current operational streamflow forecasting is prone to error due to input data uncertainties and model biases, making it difficult to accurately forecast discharge during snow melt events. Data assimilation is a technique of weighting model estimates and observations based on uncertainties that allows optimal estimation of model states. In this study, we assimilate snow water equivalent (SWE) data from the Advance Microwave Scanning Radiometer - Earth Observing System (AMSR-E) instrument into a conceptual temperature index snow model, the US National Weather Service (NWS) SNOW17 model. This model is coupled with the NWS Sacramento Soil Moisture Accounting (SAC-SMA) model, which ultimately produces stream discharge. The objective of this study is to improve the SNOW17 estimate of SWE by integrating SWE observations and uncertainties associated with meteorological forcing data within the model. For the purpose of this study, 25 km AMSR-E SWE data is used. An ensemble Kalman filter (EnKF) assimilation framework performs assimilation on a daily cycle for a 6 year period, water years 2006-2011. This method is tested on seven watersheds in the Upper Mississippi River basin that are under the forecasting jurisdiction of the NWS North Central River Forecasting Center (NCRFC). Prior to assimilation, AMSR-E data is bias corrected using data from the National Operational Hydrologic Remote Sensing Center (NOHRSC) airborne snow survey program. Discharge output from the SAC-SMA is verified using observed discharge from the outlet of each study site. Improvements in discharge are evident for five sites, in particular for high discharge magnitudes associated with snow melt runoff. Evidence points to the SNOW17 having a consistent SWE underestimation bias and error in snow melt rate. Overall results indicate that the EnKF is a viable and effective solution for integrating observations directly with operational models.