Abstract
River discharges are traditionally modeled by employing a standard power-law methodology. Recently, the Bayesian approached has successfully been applied to improve the estimates of the standard power-law. In this article, an extension to the standard power-law based on Bayesian B-splines is developed and tested on data sets from 61 different rivers. The extended model is evaluated against the standard power-law using two measures, the Deviance Information Criterion and Bayes factor. The extended model captures deviations in the data from the standard power-law but reduces to the standard power-law when that model is adequate. The standard power-law is inadequate for 26% of the rivers while the extended model provides an adequate fit in all of those cases and for the remaining 74% of the rivers the extended model and the power-law model both give adequate fit with almost identical estimates.
Similar content being viewed by others
References
Arnason S (2005) Estimating nonlinear hydrological rating curves and discharge using the Bayesian approach. Masters thesis, Faculty of Engineering, University of Iceland
Clarke R (1999) Uncertainty in the estimation of mean annual flood due to rating-curve indefinition. J Hydrol 222:185–190
Di Baldassarre G, Montanari A (2009) Uncertainty in river discharge observations: a quantitative analysis. Hydrol Earth Syst Sci 13:913–921
Gelman A, Rubin D (1992) Inference from iterative simulation using multiple sequences. Stat Sci 7(4):457–472
Gelman A, Carlin J, Stern H, Rubin D (2004) Bayesian data analysis (2nd edn.) Chapman and Hall/CRC, Boca Raton
ISO (1983) Liquid flow measurements in open channels. Handbook 16. International Standards Organization
Jeffreys H (1961) Theory of probability, 3rd edn. Oxford University Press, Oxford
Kass R, Raftery A (1995) Bayes factors and model uncertainty. J Am Stat Assoc 90:807–818
Lambie J (1978) Measurement of flow-velocity-area methods. In: Hydrometry: principles and practices. Wiley, Chichester
Lohani A, Goel N, Bhatia K (2006) Takagi–Sugeno fuzzy inference system for modeling stage–discharge relationship. J Hydrol 331:146–160
Marx B, Eilers P (2005) Multidimensional penalized signal regression. Technometrics 47:13–22
Mosley M, McKerchar A (1993) Streamflow. In: Handbook of hydrology. McGraw Hill, New York
Moyeed R, Clarke R (2005) The use of Bayesian methods for fitting rating curves, with case studies. Adv Water Resour 28:807–818
Pelletier P (1988) Uncertainties in the single determination of river discharge: a literature review. Can J Civil Eng 15:834–850
Petersen-Øverleir A (2004) Accounting for heteroscedasticity in rating curve estimates. J Hydrol 292:173–181
Petersen-Øverleir A, Reitan T (2005) Objective segmentation in compound rating curves. J Hydrol 311:188–201
Reitan T, Petersen-Øverleir A (2008) Bayesian power-law regression with a location parameter, with applications for construction of discharge rating curves. Stoch Env Res Risk Assess 22:351–365
Reitan T, Petersen-Øverleir A (2009) Bayesian methods for estimating multi-segment discharge rating curves. Stoch Env Res Risk Assess 23:627–642
Robert C (2007) The Bayesian choice: from decision-theoretic foundations to computational implementation. Springer, New York
Rue H, Held L (2005) Gaussian Markov random fields: theory and applications. Chapman & Hall/CRC, Boca Raton
Spiegelhalter D, Best N, Carlin B, van der Linde A (2002) Bayesian measures of model complexity and fit. J R Stat Soc B 64:583–639
Wasserman L (2006) All of nonparametric statistics. Springer, New York
Acknowledgments
The work presented here is based on Kristinn M. Ingimarsson’s Master’s thesis at the School of Engineering and Natural Sciences at the University of Iceland. The thesis was written under the direction of Associate Research Professor of Statistics Birgir Hrafnkelsson at the Science Institute and Professor of Environmental Engineering Sigurdur M. Gardarsson at Faculty of Civil and Environmental Engineering both at the University of Iceland and in cooperation with Arni Snorrason, Chief of the Icelandic Meteorological Office. This work was funded by the Icelandic Centre for Research (RANNIS-070241021) and has no conflict of interests. We thank the Icelandic Centre for Research for its support.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
1.1 Trace plots and the Gelman–Rubin convergence measure
Figures 9 and 10 show trace plots of the chains from the posterior simulations and plots of the evolution of the Gelman–Rubin convergence measure for Models 1 and 2, respectively.
1.2 Prior distributions
The following prior distributions are proposed for the unknown parameters.
where I(A) is such that I(A) = 1 if A is true and I(A) = 0 otherwise. In the prior distribution for λ, I is an identity matrix, M is a diagonal matrix and C is a neighbourhood matrix with constants on the first off-diagonals, other elements are equal to zero.
1.3 The Gibbs sampler
A Gibbs sampler with Metropolis–Hastings steps is used to sample from the posterior distribution. The conditional distribution of each parameter or vector of parameters in Model 2 is found and presented below. The Gibbs sampler is applied by sampling iteratively from these conditional distributions. The parameters are \(\varphi, b, c, \eta^2, b_2, c_2, \lambda, \tau^2\) and ϕ. As before, let θ be a vector containing these parameters. Terms independent of the variable being viewed are denoted by d 1, d 2 and d 3.
1.3.1 The conditional distribution of \(\varphi\)
The conditional posterior distribution of \(\varphi\) has density proportional to
The logarithm of \(p(\varphi|\hbox{rest})\) is equal to
1.3.2 The conditional distribution of b
The conditional posterior distribution of b has density proportional to
The logarithm of \( p(b|\hbox{rest})\) p(b|rest) is equal to
for 0.5 < b < 5.
1.3.3 The conditional distribution of c
The conditional posterior distribution of c has density proportional to
The logarithm of \( p(c|\hbox{rest})\) is equal to
for c < w 0.
1.3.4 The conditional distribution of η2
The conditional posterior distribution of η2 has density proportional to
Let \(\kappa = (a(w_1 - c)^b, \ldots, a(w_n - c)^b)^{{\rm T}}\) and \(g_i = (G_1(w_i), \ldots, G_L(w_i)), \) such that g i λ = ∑ L l=1 λ l G l (w i ). Let H be an n × L matrix with ith row equal to g i . Then κ + Hλ is the mean of q. Let
The logarithm of \( p(\eta^2|\hbox{rest}) \) is equal to
The above density is proportional to an inverse chi-square density, that is,
where
where s 20 = νη S 2η .
1.3.5 The conditional distribution of b 2
The conditional posterior distribution of b 2 has density proportional to
The logarithm of \( p(b_2|\hbox{rest}) \) is equal to
for 1 < b 2 < 6.
1.3.6 The conditional distribution of c 2
The conditional posterior distribution of c 2 has density proportional to
The logarithm of \( p(c_2|\hbox{rest}) \) is equal to
for c 2 < w 0.
1.3.7 The conditional distribution of λ
The conditional posterior distribution of λ has density proportional to
where the support of λ is constrained by
for all w i > w 0 and λ L = 0. Let Q λ = τ−2 M −1(I − ϕC). The logarithm of \( p(\lambda|\hbox{rest}) \) is equal to
The above density is proportional to a multivariate normal density, that is,
where
The above density is sampled under the above constraints.
1.3.8 The conditional distribution of τ2
The conditional posterior distribution of τ2 has density proportional to
The logarithm of \( p(\tau^2|\hbox{rest}) \) is equal to
The above density is proportional to an inverse chi-square density, that is,
where
1.3.9 The conditional distribution of ϕ
The conditional posterior distribution of ϕ has density proportional to
The logarithm of \( p(\phi|\hbox{rest})\) is equal to
where \(c_1, \ldots, c_L\) are the eigenvalues of C. A Metropolis–Hastings step based on the above is used to draw samples from the conditional distribution of ϕ.
Rights and permissions
About this article
Cite this article
Hrafnkelsson, B., Ingimarsson, K.M., Gardarsson, S.M. et al. Modeling discharge rating curves with Bayesian B-splines. Stoch Environ Res Risk Assess 26, 1–20 (2012). https://doi.org/10.1007/s00477-011-0526-0
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00477-011-0526-0