Characterizing and reducing equifinality by constraining a distributed catchment model with regional signatures, local observations, and process understanding

Christa Kelleher,Thorsten Wagener,Brian McGlynn

doi:10.5194/hess-21-3325-2017

Abstract

Abstract. Distributed catchment models are widely used tools for predicting hydrologic behavior. While distributed models require many parameters to describe a system, they are expected to simulate behavior that is more consistent with observed processes. However, obtaining a single set of acceptable parameters can be problematic, as parameter equifinality often results in several behavioral sets that fit observations (typically streamflow). In this study, we investigate the extent to which equifinality impacts a typical distributed modeling application. We outline a hierarchical approach to reduce the number of behavioral sets based on regional, observation-driven, and expert-knowledge-based constraints. For our application, we explore how each of these constraint classes reduced the number of behavioral parameter sets and altered distributions of spatiotemporal simulations, simulating a well-studied headwater catchment, Stringer Creek, Montana, using the distributed hydrology–soil–vegetation model (DHSVM). As a demonstrative exercise, we investigated model performance across 10 000 parameter sets. Constraints on regional signatures, the hydrograph, and two internal measurements of snow water equivalent time series reduced the number of behavioral parameter sets but still left a small number with similar goodness of fit. This subset was ultimately further reduced by incorporating pattern expectations of groundwater table depth across the catchment. Our results suggest that utilizing a hierarchical approach based on regional datasets, observations, and expert knowledge to identify behavioral parameter sets can reduce equifinality and bolster more careful application and simulation of spatiotemporal processes via distributed modeling at the catchment scale.

Highlights

The field of hydrology has been built upon the combination of field measurements and computational modeling to observe, understand, and predict hydrologic behavior (Crawford and Linsley, 1966; Beven and Kirkby, 1976, 1979; Ponce and Shetty, 1995a, b)
Across the modeling and prediction in ungauged catchment literature, we have found that constraints on model-derived hydrologic behavior generally fall into three general categories: 1. regional signatures of similarity (e.g., Yadav et al, 2007; Bloeschl et al, 2013; Hrachowitz et al, 2014), typically applied to regionalize hydrologic models for streamflow prediction in ungauged catchments; 2. objective functions or error metrics (e.g., Wagener et al, 2001; Gupta et al, 2008; van Werkhoven et al, 2008; Pfannerstill et al, 2014; Shafii and Tolson, 2015), which measure how well simulations match observations; and
While we found that redundancy did occur in this type of framework, more than half of the simulations found to be non-behavioral by constraining one metric were considered behavioral by another metric across a majority of metric combinations

Summary

Introduction

The field of hydrology has been built upon the combination of field measurements and computational modeling to observe, understand, and predict hydrologic behavior (Crawford and Linsley, 1966; Beven and Kirkby, 1976, 1979; Ponce and Shetty, 1995a, b). Towards this end, distributed, physically based models were first developed as tools to represent spatially discretized processes with physically meaningful parametric relationships (Freeze and Harlan, 1969; Beven and Kirkby, 1979; Band et al, 1991, 1993; Wigmosta et al, 1994; Refsgaard and Storm, 1995; Bixio et al, 2002; Qu, 2004; Qu and Duffy, 2007; Camporese et al, 2010; Fatichi et al, 2016). Kelleher et al.: Characterizing and reducing distributed model equifinality model parameters (see Singh and Woolhiser, 2002; Kampf and Burges, 2007; Paniconi and Putti, 2015, for a review of distributed models)

Objectives

Results

Discussion

Conclusion