Water by nature is a suitable domain for the transport of contaminants through watersheds. Evaluating the relative amounts of stored or moving water via the different components of the hydrological cycle is required for precise and strict management and planning of water resources. One of the most challenging parts of this process is the separation and quantification of baseflow from the total streamflow hydrograph. The aim of this study was to separate the storm runoff hydrograph into its components, thus being able to infer about sources and hydrological pathways of the storm runoff. The specific objectives of this study were to identify the most accurate and user-friendly streamflow partitioning method, to evaluate the accuracy of each of these methods using separately measured surface and subsurface flow data, and finally to improve available techniques or develop a more precise approach for separation of hydrograph components.

In the early stage of this study, forty different streamflow partitioning methods were reviewed and classified into three-component, analytical, empirical, graphical, geochemical and automated methods and five methods were identified as being the most relevant and least input intensive. The performance of these methods were tested against independently measured surface and subsurface flow data obtained on a field scale watershed Boughton's method produced the most consistent and accurate results. However, its accuracy depends upon the proper estimation of the end of surface runoff, and the fraction factor (α). It was demonstrated that incorporating physical and hydrologic characteristics of a watershed can significantly improve the accuracy of hydrograph separation techniques when used jointly with enhanced recession limb analysis, calibration approach, and time-discretization method. Finally, simulation of the model for different scenarios (e.g., soils, land use, etc.) was performed within the geographical information systems for a large scale watershed (Little River Watershed in Georgia). Results showed that the weighted discharge method is better than the weighted average curve number method and the modified Boughton's method because it divides a watershed into small filed scale pixels and treats each pixel separately, thus mimicking the field scale station Z conditions where the method was successfully applied.