This thesis is focused on developing a runoff estimation framework for flood prediction applications in the Namal valley, Mianwali. Prediction in the ungauged Namal watershed is challenging since it is fed entirely through rainfall by means of highly unstructured and unregulated hill torrents. The estimation system uses input from a rain gauge network physically deployed in the catchment area as part of this study. We present the sensor placement procedure which is based on the maximization of mutual information between instrumented and un-instrumented locations. We have deployed the rain stations developed by WIT-LUMS at the computed locations. Based on the real-time data, a Kriging system was developed for spatial averaging of the entire precipitation field over the watershed. Finally, simulated flow data is generated through a hydrological model of the catchment developed using the HEC-HMS platform. This data can not only be used in hydrological models of the reservoir for the prediction of extreme hydrological events such as floods but can also generate recommendations for effective gate operations.