Large Sample Basin Experiments for Hydrological Model Parameterization: Results of the Model Parameter Experiment–MOPEX. IAHS Publ. 307, 2006, 288–299.
Predicting river flow statistics at ungauged
locations—a hydrostochastic approach
Eric Sauquet1, Lars Gottschalk2, Irina Krasovskaia2 & Etienne
Leblois1
1 Cemagref, Hydrology - Hydraulics Research Unit, 3 bis quai Chauveau CP220, F-69336 Lyon cedex 09, France
eric.sauquet@cemagref.fr
2 Department of
Geosciences, University of Oslo, PO Box 1022 Blindern, N-0315 Oslo, Norway
Abstract The specific objective of this paper is to provide an overview of the collaborative projects between the University of Oslo (Norway) and Cemagref (France). The research undertaken concentrated on the development of methods for estimating statistical properties of the water balance components (especially runoff) at ungauged locations. Probability and Statistical Theory is used to estimate runoff characteristics related to various scales in time and space. A river basin was studied as a system consisting of interacting entities without splitting it into these entities—looking for patterns rather than isolated steps of causality, i.e. letting the data speak for themselves. Links between the water balance statistics and the way they develop in time and/or in space, i.e. along the river network, are analysed and these dependences are introduced explicitly in interpolation procedures. Thus the approach preserves the statistical properties in terms of covariances and semi-variograms at a basin scale as well as satisfying water balance constraints. A review of the basic concepts is given first. Second, applications of the concepts are presented with examples of their application to different extensive datasets (France and Costa Rica) and for different hydrological variables. Mean annual and monthly discharges and flow duration curves are investigated in this article.
key words empirical orthogonal function; flow duration curve; runoff; scale;
stochastic interpolation