Investigating the relationship between hydrologic model parameters and physical catchment metrics for improved modeling in data-sparse regions
Abstract
In regions with sparse data, hydrologic modelers often endeavor to transfer information from longer-term gauged catchments to those with limited data. In this approach, it is assumed that these gauged ';surrogates' can provide useful information for those ungauged catchments that are hydrologically similar. One recent method aims to pool catchments with similar hydrologic behavior so that models may be more convincingly applied to catchments without detailed observations. An ongoing concern, however, is how to identify catchments that behave similarly in terms of hydrologic processes and thus classify catchments in terms of their modeled behavior. In this study, we investigate the complex relationship between physical catchment characteristics, hydrologic signatures, and optimized hydrologic models for regions with sparse data. We make use of a data set of over 150 catchments located in southeast Australia with basic climatic and hydrologic time series and limited information on physical catchment characteristics. A conceptual rainfall-runoff model is calibrated for each of the catchments and hierarchical clustering is performed to link catchments based on their calibrated model parameters. We then aim to isolate the physical and spatial metrics that are common to each member of a given cluster with the ultimate goal of providing insight to the selection of gauged surrogates for ungauged watersheds. A Permutational Multivariate Analysis of Variance (perMANOVA) is performed to determine if significant differences exist between clusters according to certain physical and climatic catchment descriptors. We further analyze the data using a classification tree to determine the extent to which cluster membership can be predicted by basic catchment descriptors. Our results show support for the 'surrogate' technique for hydrologic regionalization by demonstrating that the clusters, though built using calibrated model parameters, are related to clear differences in the catchment descriptors.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFM.H53E1464M
- Keywords:
-
- 1847 HYDROLOGY Modeling;
- 1874 HYDROLOGY Ungaged basins