To main content

Evaluation of different parameterizations of the spatial heterogeneity of subsurface storage capacity for hourly runoff simulation in boreal mountainous watershed

Abstract

Identification of proper parameterizations of spatial heterogeneity is required for precipitation-runoff models. However, relevant studies with a specific aim at hourly runoff simulation in boreal mountainous catchments are not common.

We conducted calibration and evaluation of hourly runoff simulation in a boreal mountainous watershed based on six different parameterizations of the spatial heterogeneity of subsurface storage capacity for a semi-distributed (subcatchments hereafter called elements) and distributed (1x1 km2 grid) setup. We evaluated representation of element-to-element, grid-to-grid, and probabilistic subcatchment/subbasin, subelement and subgrid heterogeneities.

The parameterization cases satisfactorily reproduced the streamflow hydrographs with Nash-Sutcliffe efficiency values for the calibration and validation periods up to 0.84 and 0.86 respectively, and similarly for the log-transformed streamflow up to 0.85 and 0.90. The parameterizations reproduced the flow duration curves, but predictive reliability in terms of quantile-quantile (Q-Q) plots indicated marked over and under predictions. The simple and parsimonious parameterizations with no subelement or no subgrid heterogeneities provided equivalent simulation performance compared to the more complex cases. The results indicated that (i) identification of parameterizations require measurements from denser precipitation stations than what is required for acceptable calibration of the precipitation-streamflow relationships, (ii) there is challenges in the identification of parameterizations based on only calibration to catchment integrated streamflow observations and (iii) a potential preference for the simple and parsimonious parameterizations for operational forecast contingent on their equivalent simulation performance for the available input data. In addition, the effects of non-identifiability of parameters (interactions and equifinality) can contribute to the non-identifiability of the parameterizations.

Category

Academic article

Client

  • Research Council of Norway (RCN) / 193818

Language

English

Author(s)

  • Teklu Tesfaye Hailegeorgis
  • Knut Alfredsen
  • Yisak Sultan Abdella
  • Sjur Kolberg

Affiliation

  • Norwegian University of Science and Technology
  • SINTEF Energy Research / Energisystemer

Year

2015

Published in

Journal of Hydrology

ISSN

0022-1694

Publisher

Elsevier

Volume

522

Page(s)

522 - 533

View this publication at Cristin