The GIS implementation is designed to be a user interactive method requiring three input grid datasets. The three required data inputs are catchment boundaries, rainfall, and landuse (current forest cover).
It is then easy to predict the mean annual water balance (stream flow) of the catchment. The GIS implementation is also designed to enable scenario testing so that it can predict changes in water yield as a result of changes in landuse.
you can use Muskingum River Routing ( streamflow routing and superposition-after-routing means which means Streamflow is routed from external upstream node to the internal downstream junction node.
Have a look on the free matlab toolbox topotoolbox. It provides very nice tools for what regards hydrology.
Difficult to say definitively which is the best remote sensing sensor. With wide range spetral bands in the visible-infrared range (e.g., Landasat 2, modis) you can roughly estimate soli water content (Brocca, L., et al. "Soil moisture estimation in alpine catchments through modeling and satellite observations." Vadose Zone Journal 12.3 (2013).
EUMETSAT has products, where evapotranspiration (ET) is calculated. For example, I use one of these to integrate ET values at daly, 10-day or month periods. Example is on the pictures.
Drawback of the product is a fact, that algorithm does not take into account ice evaporation.