Modelling hydrologic responses in a small forested catchment (Panola Mountain, Georgia, USA): a comparison of the original and a new dynamic TOPMODEL
Version of Record online: 23 JAN 2003
Copyright © 2003 John Wiley & Sons, Ltd.
Special Issue: Runoff Generation and Implications for River Basin Modelling
Volume 17, Issue 2, pages 345–362, 15 February 2003
How to Cite
Peters, N. E., Freer, J. and Beven, K. (2003), Modelling hydrologic responses in a small forested catchment (Panola Mountain, Georgia, USA): a comparison of the original and a new dynamic TOPMODEL. Hydrol. Process., 17: 345–362. doi: 10.1002/hyp.1128
- Issue online: 23 JAN 2003
- Version of Record online: 23 JAN 2003
- Manuscript Accepted: 2 JAN 2002
- Manuscript Received: 31 JAN 2001
- US National Science Foundation. Grant Number: EAR 9743311
- United Kingdom National Environmental Research Council. Grant Number: GR3/11450
- distributed hydrological model;
- streamflow generation;
Preliminary modelling results for a new version of the rainfall-runoff model TOPMODEL, dynamic TOPMODEL, are compared with those of the original TOPMODEL formulation for predicting streamflow at the Panola Mountain Research Watershed, Georgia. Dynamic TOPMODEL uses a kinematic wave routing of subsurface flow, which allows for dynamically variable upslope contributing areas, while retaining the concept of hydrological similarity to increase computational efficiency. Model performance in predicting discharge was assessed for the original TOPMODEL and for one landscape unit (LU) and three LU versions of the dynamic TOPMODEL (a bare rock area, hillslope with regolith <1 m, and a riparian zone with regolith ⩽5 m). All simulations used a 30 min time step for each of three water years. Each 1-LU model underpredicted the peak streamflow, and generally overpredicted recession streamflow during wet periods and underpredicted during dry periods. The difference between predicted recession streamflow generally was less for the dynamic TOPMODEL and smallest for the 3-LU model. Bayesian combination of results for different water years within the GLUE methodology left no behavioural original or 1-LU dynamic models and only 168 (of 96 000 sample parameter sets) for the 3-LU model. The efficiency for the streamflow prediction of the best 3-LU model was 0·83 for an individual year, but the results suggest that further improvements could be made. Copyright © 2003 John Wiley & Sons, Ltd.