Flood became more frequent and devastating disaster in recent world. Flood forecasting is an essential tool for flood mitigation strategies. The Koshi River is one of the major tributaries of the Ganges. Before joining the Ganges River, almost every year it floods the downstream areas. As upstream area in Himalayan range, hydro-meteorological data is very scarce in that region. Using remote sensing technology to abstract hydro-meteorological data is possible to acquire and this technology is emerging day by day. The Koshi River has a number of tributaries and all of them have originated from Himalayan range. This river basin has high elevation difference. High elevated upstream region prevails an arctic climate whereas low elevated downstream region exhibits tropical climate. Glacier are a common feature in the upstream area. To develop flood forecasting model, data acquisition was done from different sources. To estimate flow in real time, a technique was developed using passive microwave AMSR-E image. This technique exhibits about 75% correlation between simulated and observed flow with 0.0092 m/km ground slope or lower. Above this slope this microwave AMSR-E approach technique does not show acceptable correlation, because that region is denoted as mountainous region. Rivers in mountainous region are comparatively narrow and there is no flood plain. To overcome this problem, another technique was developed using water surface reflectance of LANDSAT imagery. The principle of water surface reflectance approach for remote flow estimation is the attenuation attribute of spectrum in water. This attenuation attribute can be correlate with observed flow. LANDSAT images are not available in daily basis and also interfered by cloud cover in monsoon. Simulated data shows acceptable error with observed flow. Comparison among observed flow, estimated flow and flow from NAM model was performed and results shows acceptable correlation.Hydrological model was development using Aphrodite rainfall appears more suitable for flood forecasting. To acquire recent temperature and evaporation data, MODIS LST and temperature vs evaporation correlation was used respectively. Data acquisition for hydrodynamic model was done using DHM flow data and DEM data. As no historical water level information was not available, flow forecasting was performed. Using GFS or ECMWF rainfall forecast, flow forecasting exhibits random results. Using pre-monsoon and monsoon flow to test the FF model shows that, flow forecast is in acceptable limit upto 2-4 day. After 4 days flow forecast has high error. This FF model is has capability to work with recent dataset and forecast in real time.