Snow cover estimation is a principal source of error for spring streamflow simulations in Québec, Canada. Optical and near infrared remote sensing can improve snow cover area (SCA) estimation due to high spatial resolution but is limited by cloud cover and incoming solar radiation. Passive microwave remote sensing is complementary by its near-transparence to cloud cover and independence to incoming solar radiation, but is limited by its coarse spatial resolution. The study aims to create an improved SCA product from blended passive microwave (AMSR-E daily L3 Brightness Temperature) and optical (MODIS Terra and Aqua daily snow cover L3) remote sensing data in order to improve estimation of river streamflow caused by snowmelt with Québec's operational MOHYSE hydrological model through direct-insertion of the blended SCA product in a coupled snowmelt module (SPH-AV). SCA estimated from AMSR-E data is first compared with SCA estimated with MODIS, as well as with in situ snow depth measurements. Results show good agreement (+95%) between AMSR-E-derived and MODIS-derived SCA products in spring but comparisons with Environment Canada ground stations and SCA derived from Advanced Very High Resolution Radiometer (AVHRR) data show lesser agreements (83 % and 74% respectively). Results also show that AMSR-E generally underestimates SCA. Assimilating the blended snow product in SPH-AV coupled with MOHYSE yields significant improvement of simulated streamflow for the aux Écorces et au Saumon rivers overall when compared with simulations with no update during thaw events. These improvements are similar to results driven by biweekly ground data. Assimilation of remotely-sensed passive microwave data was also found to have little positive impact on springflood forecast due to the difficulty in differentiating melting snow from snow-free surfaces. Considering the direct-insertion and Newtonian nudging assimilation methods, the study also shows the latter method to be superior to the former, notably when assimilating noisy data.