Keywords
hydrological modelling, regionalisation, equifinality, ungauged basins
Start Date
1-7-2004 12:00 AM
Abstract
Development of methodologies to achieve a priori parameter estimation of hydrological models is fundamental in ungauged basins, to reduce the number of parameters to be calibrated or to obtain parameter values where calibration is not possible. This work shows that conceptual rainfall-runoff models can be applied to ungauged watersheds by developing relationships between model parameters and watershed characteristics. In fact, the calibration of MEDOR, a four parameter daily lumped conceptual rainfall-runoff model specific for the Mediterranean catchments, is affected by the equifinality issue. Systematic scanning of the Nash criterion objective function demonstrates that a Production Equifinality Relationship (PER) exists between loss function parameters. This basin-specific relation can be determined using the annual balance of rainfall-runoff and daily rainfall data generated by a stochastic model calibrated for the region. Moreover, the analysis shows the importance of the stochastic structure of rainfall in the calibration of MEDOR. Thus, the parameters cannot be determined solely from the physical properties of the basin. Coupled to a stochastic model of rainfall of a given region, MEDOR generates equifinality relations between runoff coefficients (CE) defining a surface in the parameters space. Several large areas have been identified in the Mediterranean region having a single reference CE surface (e.g., East coast of the Mediterranean Sea). The runoff coefficient of a given watershed located in one of these areas constrains the specific equifinality relation. In a case study, the PER of two Lebanese watersheds were determined using the same single reference CE surface and spatial soil depth information. Thus, it is possible to evaluate the predictive uncertainty for the streamflow at the catchments outlet.
Parameter Estimation of a Conceptual Rainfall Runoff Model and Application to Mediterranean Catchments
Development of methodologies to achieve a priori parameter estimation of hydrological models is fundamental in ungauged basins, to reduce the number of parameters to be calibrated or to obtain parameter values where calibration is not possible. This work shows that conceptual rainfall-runoff models can be applied to ungauged watersheds by developing relationships between model parameters and watershed characteristics. In fact, the calibration of MEDOR, a four parameter daily lumped conceptual rainfall-runoff model specific for the Mediterranean catchments, is affected by the equifinality issue. Systematic scanning of the Nash criterion objective function demonstrates that a Production Equifinality Relationship (PER) exists between loss function parameters. This basin-specific relation can be determined using the annual balance of rainfall-runoff and daily rainfall data generated by a stochastic model calibrated for the region. Moreover, the analysis shows the importance of the stochastic structure of rainfall in the calibration of MEDOR. Thus, the parameters cannot be determined solely from the physical properties of the basin. Coupled to a stochastic model of rainfall of a given region, MEDOR generates equifinality relations between runoff coefficients (CE) defining a surface in the parameters space. Several large areas have been identified in the Mediterranean region having a single reference CE surface (e.g., East coast of the Mediterranean Sea). The runoff coefficient of a given watershed located in one of these areas constrains the specific equifinality relation. In a case study, the PER of two Lebanese watersheds were determined using the same single reference CE surface and spatial soil depth information. Thus, it is possible to evaluate the predictive uncertainty for the streamflow at the catchments outlet.
