Non-Linear Signal Processing of Rainfall- Runoff Process Using Data-Based Mechanistic Modeling (DBM) Case Study: Khersan Basin

There were many hydrological researches focused on dynamic and linear modeling of rainfall-runoff process. Conversion of rainfall data to runoff consists of nonlinear complex relationships which are resulted from interactions of different sets of hydrological process. The stochastic modeling is therefore seems to be more sensible in this estimate than deterministic ones. In this research observed runoff is modeled against total rainfall. This will mainly avoid misleading theories in breaking the rainfall and runoff time series into the excess rainfall and the direct runoff time series,. A Transfer Function (TF) model with single input and single output variable (SISO) is used in this research. This function is transferred to the state space equations. The stochastic Data-Based Mechanistic modeling (DBM) method relying upon recursive Kalman filtering algorithm is then used to identify the non-linear relationship between rainfall and runoff. This approach is applied to the Khersan sub basin in the Great Karun catchment south western Iran. The relation between the calibrated parameters and the routine characteristics of the basin flow showed a probable parallel structure of flow routine in this sub basin. Finally the sensitivity analysis is performed using the Monte Carlo Simulation (MCS) in order to quantify the reliability of the model.