A total of nine two-month, free-running simulations are performed for August and September of 2008, 2009, and 2010. One set of (three) simulations acts as a control run, using ECMWF ERA-Interim for initial conditions and boundary conditions. An identical set of (three) simulations is performed, but a dry plateau scenario (Dplat) is imposed, where soil moisture over the Deccan Plateau is initialized as the lowest volumetric soil moisture content over a 30-year climatology. Another identical set of (three) simulations is also performed, but a wet plateau scenario (Wplat) is imposed, where soil moisture over the Deccan Plateau is initialized as the highest volumetric soil moisture content over a 30-year climatology. We evaluate precipitation produced by the model against TRMM PR rainfall climatology as well as surface observations from a unique rain gauge network centered over the Western Ghats. Precipitation, meteorological parameters, and land surface variables show the least sensitivity to soil moisture initialization in regions upstream of the Western Ghats and immediate to the mountains themselves. Over the plateau, however, high sensitivity of precipitation to soil moisture initialization is observed for 5-10 days into the simulation. Notable sensitivity exists through the duration of the two-month simulations, especially during break periods, suggesting a positive feedback primarily between dry soil conditions and decreased precipitation over the plateau.