Effect of Spatial Scale on Hydrologic Modeling in a Headwater Catchment

In this study, two hydrologic models were applied to the mountainous Back Creek catchment, located in the headwaters of the Roanoke River in Southwest Virginia. The two models employed were HEC-1, an event based lumped model, and TOPMODEL, a continuous semi-distributed model. These models were used to investigate (a) the issue of spatial scale in hydrologic modeling, and (b) two approaches to modeling, continuous versus event based. Two HEC-1 models were developed with a different number of subareas in each. The hydrographs generated by each HEC-1 model for a number of large rainfall events were analyzed visually and statistically. No observable improvement resulted from increasing the number of subareas in the HEC-1 models from 20 to 81.

TOPMODEL was applied to the same watershed using a series of different size grid cells. The first step in applying TOPMODEL to a watershed involves GIS analysis which results in a raster grid of elevations used for the calculation of the topographic index, ln(a/tan b). The hydrographs generated by TOPMODEL with each grid cell size were compared in order to assess the sensitivity of TOPMODEL hydrographs to grid cell size. An increase in grid cell size from 15 to 120 meters resulted in increased values of the watershed mean of the topographic index. However, hydrographs generated by TOPMODEL were completely unaffected by this increase in the topographic index. Analyses were also performed to determine the sensitivity of TOPMODEL hydrographs to several model parameters. It was determined that the parameters that had the greatest effect on hydrographs generated by TOPMODEL were the m and ln(To) parameters.

The modeling performances of the event based HEC-1 and the continuous TOPMODEL were analyzed and compared visually and statistically for a number of large storms. The limited number of storms used to compare HEC-1 and TOPMODEL makes it difficult to determine definitively which model simulates large storms better. It does appear that perhaps HEC-1 is slightly superior in that regard. TOPMODEL was also executed as an event based model for two single events and the resulting hydrographs were compared to the HEC-1 and continuous TOPMODEL results. Both HEC-1 and TOPMODEL (when used as a continuous model) simulate large storms better than TOPMODEL (when used as an event based model).