Search

Summary The BTOPMC (Blockwise use of TOPMODEL with Muskingum–Cunge routing) model was tested for spatial and temporal transferability of the calibrated model parameters to predict various runoff components (runoff volumes, peak flows and base flows) in different physiographic regions of Nepal. To evaluate the spatial transferability, the model was calibrated and validated using available time series data in six catchments. Results show that the model performed well in simulating various runoff components and capturing the seasonal and annual variations in runoff. The model parameters of donor catchments were selected based on the similarity index and applied in other two data limited catchments using the regionalisation scheme. Results show that the model could reproduce the runoff volume, base flows, seasonal and annual variations in runoff when the model parameters were transferred spatially. To evaluate temporal transferability, two different scenarios were tested in a river catchment, Bagmati river. The first scenario used three data sets: 1994–1997 for calibration, 1998–1999 for validation, and 2000–2002 data for evaluation. The second scenario also used three data sets: 2000–2002 for calibration, 2003 for validation and 1994–1997 for evaluating transferability of model parameters. Only a marginal loss in model performance was observed when calibrated parameters were transferred from one period to another period. This study therefore suggests that the calibrated parameters of BTOPMC model can be spatially and temporally transferred within river catchments located in similar physiographic regions and especially in those catchments where data is sparse. Such regionalized models can be used for hydrological simulations to study the impacts of land use and climate changes.

Summary One of key inputs to hydrological modeling is the potential evapotranspiration, either from the interception (PET 0 ) or from the soil water of root zone (PET). The Shuttleworth–Wallace (S–W) model is developed for their estimation. In this parameterization, neither experimental measurement nor calibration is introduced. Based on IGBP land cover classification, the typical thresholds of vegetation parameters are drawn from the literature. The spatial and temporal variation of vegetation LAI is derived from the composite NOAA-AVHRR NDVI using the SiB2 method. The CRU database supplies with the required meteorological data. They are all publicly available. The developed S–W model is applicable at the global scale, particularly to the data-poor or ungauged large basins. Using the century monthly time series of CRU TS 2.0 and the monthly composite NOAA-AVHRR NDVI from 1981 to 2000, annual PET is estimated 1354 mm over the Mekong River basin, spatially distributed strikingly non-uniformly from 300 to 2040 mm, and seasonally changed significantly with LAI. By replacing the monthly with the 10-day composite NDVI and the albedo of 0.10 with 0.15 for substrate soil surface, annual PET relatively decreases less than 4% and 1.7%, respectively over the whole basin. The correlation with pan evaporation ( E pan ) is quite scattered but grouped with the vegetation types and consistent with a rough ratio as reported in the literature. In contrast, the PET and the reference evapotranspiration (RET) are vegetation-type-dependently correlated very well. The PET 0 is estimated 1.63 times of PET in average over the whole basin. The application of BTOPMC model shows that the derived LAI, PET 0 and PET behave very well in the distributed hydrological modeling.

This paper focuses on parameter regionalization for the block-wise use of the TOPMODEL with the Muskingum–Cunge routing method (BTOPMC). Pre-determined quantitative relationships with physical basin characteristics are used to reduce uncertainty in parameter estimation and enable prediction in ungauged basins. BTOPMC parameter values, which have previously resulted in efficient model performance, are applied to ten Model Parameter Estimation Experiment (MOPEX) basins without any parameter adjustment to test model performance using ‘blind’ a priori parameter estimates. It is quantitatively shown that parameters can not be satisfactorily estimated a priori without parameter adjustment, either through calibration or transfer functions linking parameters to physical basin features, referred to as ‘improved’ a priori parameter estimation. To compare model performance using different parameter estimation techniques, BTOPMC parameters are re-estimated by automatic optimization. As anticipated, model performance improves significantly when these optimized parameters are used. Good correlation of model performance is also observed between the 19-year validation and 20-year calibration results when optimized parameters are used. This indicates that BTOPMC's predictive accuracy depends strongly on calibration accuracy, implying that well-optimized parameter values can ensure the accuracy of model predictions. It appears reasonable to use optimized parameter values, selected from successful modelling applications, to make transfer functions which ‘improve’ a priori parameter estimates. The ‘improved’ a priori parameter estimation is achieved through multiple linear regression analysis relating BTOPMC parameters to soil and vegetation types within the basin being modelled. The results suggest that the preliminary parameter transfer functions are encouraging and that ‘improved’ a priori parameter estimation techniques have the potential to reduce parameter uncertainty and enable prediction in ungauged basins. To improve parameter estimation and predictive accuracy of the model, further research to determine more accurate values, and their relationship with physical basin characteristics, is needed.