A methodology for preserving channel flow networks and connectivity patterns in large-scale distributed hydrological models

摘要

Physiographic data are often used to parameterize hydrological models and, in the past, physiographic parameters have often been derived manually. However, this can be a lengthy and unreliable process, particularly for application to a gridded hydrological or atmospheric model applied to large or continental-scale basins. An important attribute of gridded models is drainage direction. Current methods that determine drainage directions for large or continental-scale basins, by general circulation models (GCMs), route flow using lowest neighbour algorithms. These methods, however, do not reflect the hydrology of the basin subunit. This paper proposes a method of parameterizing hydrological models with physiographic data using the Arclnfo macro language to create an interface between the Topographic Parameterization (TOPAZ) software and the WATFLOOD hydrological model. The interface uses output raster data created by TOPAZ (i.e. drainage identification) to supply physiographic parameters required by WATFLOOD. The interface (WATPAZ) is an expert system based on a manual method of deriving parameters for the WATFLOOD distributed model. The WATPAZ interface uses grouped response units to subdivide the watershed. This allows large drainage basins to be subdivided at a scale that allows computational efficiency while preserving the hydrological variability of the watershed. To test whether the WATPAZ method improves the current GCM methodology for determining drainage directions, WATPAZ is applied on a local basin (Wolf Creek) a regional-scale basin, (Athabasca) and a continental-scale basin (Mackenzie). An examination of flow directions derived from this new method with current GCM methods is carried out. The results indicate that a substantial improvement is made to flow routing within the basin using the channel network to determine drainage directions for each segment.