Abstract
Accurate rainfall runoff simulation is essential for river basin management in monsoon dominated regions where hydrological responses are highly nonlinear and uncertain. This study develops a Monte Carlo based framework to evaluate predictive uncertainty and parameter sensitivity in rainfall runoff modelling for the Wainganga River Basin a tributary of the Godavari River India. A GIS integrated lumped Soil Moisture Accounting (SMA) model was implemented at a daily time scale. Model calibration was conducted for 2001–2012 followed by validation for 2013–2024 using observed stream flow data. The model demonstrated satisfactory performance achieving Nash Sutcliffe Efficiency (NSE) values of 0.78 during calibration and 0.72 during validation with corresponding RMSE values of 42.6 m³/s and 48.3 m³/s respectively. Parameter uncertainty was explored through extensive Monte Carlo sampling within physically plausible ranges. Global sensitivity analysis indicated that Muskingum routing parameters (K and X) exert the strongest influence on simulated discharge whereas soil moisture storage parameters showed comparatively lower sensitivity. Predictive uncertainty was quantified using likelihood weighted ensembles, producing 5–95th percentile discharge envelopes that encompassed approximately 78% of observed flows increasing to about 84% during the monsoon period. The results demonstrate that incorporating uncertainty analysis enhances confidence in rainfall runoff simulations and supports more reliable flood risk and water resources management in data-scarce monsoon basins.