Parameter-dependent convergence bounds and complexity measure for a class of conceptual hydrological models

We provide analytical bounds on convergence rates for a class of hydrologic models and consequently derive a complexity measure based on the Vapnik-Chervonenkis (VC) generalization theory. The class of hydrologic models is a spatially explicit interconnected set of linear reservoirs with the aim of representing globally nonlinear hydrologic behavior by locally linear models. Here, by convergence rate, we mean convergence of the empirical risk to the expected risk. The derived measure of complexity measures a model's propensity to overfit data. We explore how data finiteness can affect model selection for this class of hydrologic model and provide theoretical results on how model performance on a finite sample converges to its expected performance as data size approaches infinity. These bounds can then be used for model selection, as the bounds provide a tradeoff between model complexity and model performance on finite data. The convergence bounds for the considered hydrologic models depend on the magnitude of their parameters, which are the recession parameters of constituting linear reservoirs. Further, the complexity of hydrologic models not only varies with the magnitude of their parameters but also depends on the network structure of the models (in terms of the spatial heterogeneity of parameters and the nature of hydrologic connectivity). © IWA Publishing 2012.

[1]  Murugesu Sivapalan,et al.  Power law catchment‐scale recessions arising from heterogeneous linear small‐scale dynamics , 2009 .

[2]  Noga Alon,et al.  Scale-sensitive dimensions, uniform convergence, and learnability , 1997, JACM.

[3]  David Haussler,et al.  Equivalence of models for polynomial learnability , 1988, COLT '88.

[4]  Leslie G. Valiant,et al.  A theory of the learnable , 1984, CACM.

[5]  Philip W. L. Fong.  A Quantitative Study of Hypothesis Selection , 1995, ICML.

[6]  David Haussler,et al.  Occam's Razor , 1987, Inf. Process. Lett..

[7]  R. Govindaraju,et al.  Applicability of linearized Boussinesq equation for modeling bank storage under uncertain aquifer parameters , 1994 .

[8]  Luis A. Bastidas,et al.  Complexity‐based robust hydrologic prediction , 2009 .

[9]  Massimiliano Pontil,et al.  A note on different covering numbers in learning theory , 2003, J. Complex..

[10]  Dawei Han,et al.  Flood forecasting using support vector machines , 2007 .

[11]  Vladimir N. Vapnik,et al.  The Nature of Statistical Learning Theory , 2000, Statistics for Engineering and Information Science.

[12]  Thomas Meixner,et al.  A global and efficient multi-objective auto-calibration and uncertainty estimation method for water quality catchment models , 2007 .

[13]  P. Bartlett,et al.  The complexity of model classes, and smoothing noisy data 1 1 An earlier version of this paper was p , 1998 .

[14]  S. P. Neuman,et al.  Sensitivity analysis and assessment of prior model probabilities in MLBMA with application to unsaturated fractured tuff , 2005 .

[15]  A. Sahuquillo,et al.  An efficient conceptual model to simulate surface water body‐aquifer interaction in conjunctive use management models , 2007 .

[16]  S. Sorooshian,et al.  Uniqueness and observability of conceptual rainfall‐runoff model parameters: The percolation process examined , 1983 .

[17]  W. Brutsaert,et al.  Research Note On the first and second linearization of the Boussinesq equation , 1966 .

[18]  Edward C. Waymire,et al.  A Large Deviation Rate and Central Limit Theorem for Horton Ratios , 1991, SIAM J. Discret. Math..

[19]  S. P. Neuman,et al.  On model selection criteria in multimodel analysis , 2007 .

[20]  A. Bárdossy,et al.  A two steps disaggregation method for highly seasonal monthly rainfall , 2002 .

[21]  S. Sorooshian,et al.  Effective and efficient algorithm for multiobjective optimization of hydrologic models , 2003 .

[22]  S. Sorooshian,et al.  Effective and efficient global optimization for conceptual rainfall‐runoff models , 1992 .

[23]  C. Paniconi,et al.  Hillslope‐storage Boussinesq model for subsurface flow and variable source areas along complex hillslopes: 2. Intercomparison with a three‐dimensional Richards equation model , 2003 .

[24]  S. Sorooshian,et al.  A Shuffled Complex Evolution Metropolis algorithm for optimization and uncertainty assessment of hydrologic model parameters , 2002 .

[25]  Felipe Cucker,et al.  On the mathematical foundations of learning , 2001 .

[26]  Robert E. Schapire,et al.  Efficient distribution-free learning of probabilistic concepts , 1990, Proceedings [1990] 31st Annual Symposium on Foundations of Computer Science.

[27]  G. Lugosi,et al.  Adaptive Model Selection Using Empirical Complexities , 1998 .

[28]  Tammo S. Steenhuis,et al.  Determining watershed response in data poor environments with remotely sensed small reservoirs as runoff gauges , 2009 .

[29]  John Ockendon,et al.  “Waiting-time” Solutions of a Nonlinear Diffusion Equation , 1982 .

[30]  W. Hoeffding.  Probability Inequalities for sums of Bounded Random Variables , 1963 .

[31]  David Haussler,et al.  Learnability and the Vapnik-Chervonenkis dimension , 1989, JACM.

[32]  Hubert H. G. Savenije,et al.  Model complexity control for hydrologic prediction , 2008 .

[33]  S. Uhlenbrook,et al.  Hydrological process representation at the meso-scale: the potential of a distributed, conceptual catchment model , 2004 .

[34]  J. Freer,et al.  Consistency between hydrological models and field observations: linking processes at the hillslope scale to hydrological responses at the watershed scale , 2009 .

[35]  Peter L. Bartlett,et al.  The Sample Complexity of Pattern Classification with Neural Networks: The Size of the Weights is More Important than the Size of the Network , 1998, IEEE Trans. Inf. Theory.

[36]  Ron Meir,et al.  Nonparametric Time Series Prediction Through Adaptive Model Selection , 2000, Machine Learning.

[37]  Gábor Lugosi,et al.  Concentration Inequalities , 2008, COLT.

[38]  A. Siamj.,et al.  ON AN INVERSE DIFFUSION PROBLEM , 1997 .

[39]  Vladimir Vapnik,et al.  Chervonenkis: On the uniform convergence of relative frequencies of events to their probabilities , 1971 .

[40]  Soroosh Sorooshian,et al.  Toward improved calibration of hydrologic models: Multiple and noncommensurable measures of information , 1998 .

[41]  Keith Beven,et al.  The future of distributed models: model calibration and uncertainty prediction. , 1992 .