Multi‐variable parameter estimation to increase confidence in hydrological modelling

Abstract

Abstract:The expanding use and increased complexity of hydrological runoff models has given rise to a concern about overparameterization and risks for compensating errors. One proposed way out is the calibration and validation against additional observations, such as snow, soil moisture, groundwater or water quality. A general problem, however, when calibrating the model against more than one variable is the strategy for parameter estimation. The most straightforward method is to calibrate the model components sequentially. Recent results show that in this way the model may be locked up in a parameter setting, which is good enough for one variable but excludes proper simulation of other variables. This is particularly the case for water quality modelling, where a small compromise in terms of runoff simulation may lead to dramatically better simulations of water quality. This calls for an integrated model calibration procedure with a criterion that integrates more aspects on model performance than just river runoff.The use of multi-variable parameter estimation and internal control of the HBV hydrological model is discussed and highlighted by two case studies. The first example is from a forested basin in northern Sweden and the second one is from an agricultural basin in the south of the country. A new calibration strategy, which is integrated rather than sequential, is proposed and tested. It is concluded that comparison of model results with more measurements than only runoff can lead to increased confidence in the physical relevance of the model, and that the new calibration strategy can be useful for further model development.

Multi-criteria assessment of the Representative Elementary Watershed approach on the Donga catchment (Benin) using a downward approach of model complexity

“…In order to assess the model structure, the evaluation of a distributed model representing various processes should be done on each hydrological compartment it represents and at various scales (Refsgaard, 1997;Michaud and Sorooshian;1994;Bergström et al, 2002;Fortin et al 2001, Anderton et al 2002. As a single and integrative variable (i.e.…”

Multi-site calibration, validation, and sensitivity analysis of the MIKE SHE Model for a large watershed in northern China

“…Ambroise et al, 1995;Andersen et al, 2001;Khu et al, 2008;Dai et al, 2010). Bergström et al (2002) suggested that the model calibrated against more measured internal variables rather than streamflow only can greatly increase confidence in the physical relevance of the model. Vázquez et al (2008) adopted a multi-criteria protocol which included statistical, analytical and visual criteria to calibrate the model.…”

Modeling Glacier Mass Balance and Runoff in the Koxkar River Basin on the South Slope of the Tianshan Mountains, China, from 1959 to 2009

“…The HBV light is a conceptual mode which is based on the HBV model [23]. It was coupled with a more detailed snow and glacier melt subroutine, employing the degree-day approach [24].…”

“…Rainfall and snowmelt (P) are divided into water filling the soil box and groundwater recharge depending on the relation between water content of the soil box (SM, in mm) and its largest value (FC, in mm) as shown in Equation (4). Actual evaporation from the soil box equals the potential evaporation if SM/FC is above LP (the threshold of Soil water content and Field water holding capacity), whereas a linear reduction is used when SM/FC is below LP, as shown in Equation (5) [23]. Groundwater recharge is added to the upper groundwater box (SUZ, in mm).…”