Book Review Review of Advances in Data-Based Approaches for Hydrologic Modeling and Forecasting by B. Sivakumar and R. Berndtsson World Scientific Publishing, Hackensack, NJ; 2010; Price: $122; ISBN 13-978-981-4307-97-0; 519 pp. Vijay P. Singh, F.ASCE Caroline and William N. Lehrer Distinguished Chair in Water Engineering, Professor of Civil and Environmental Engineering, and Professor of Bio- logical and Agricultural Engineering, Dept. of Biological and Agricultural Engineering, Texas A&M Univ., Scoates Hall, 2117 TAMU, College Sta- tion, TX 77843-2117. E-mail: vsingh@tamu.edu In the preface the authors write, “…there has been, in recent years, an exponential increase in the number of scientific approaches and their applications for hydrologic modeling and forecasting. Among these, the so-called ‘data-based’ or ‘data-driven’ approaches have become particularly popular…none of the existing books, it is fair to say, is adequate enough to learn about the overall progress and the state-of-the-art of data-based approaches in hydrologic model- ing and forecasting…an attempt is made in this book to present a comprehensive account of the advances in data-based approaches for modeling and forecasting hydrologic systems and processes. ” This book responds to the need expressed in the preface, providing a clear and balanced treatment of some of the major data-based approaches, encompassing 10 chapters written by hydrologists and water resources engineers who are well known for their con- tributions. Providing the background and organization of the book, Chapter 1, written by B. Sivakumar and R. Berndtsson, sets the stage for what is to come in the ensuing chapters. Stochastic methods for modeling precipitation and streamflow, written by B. Rajagopalan, J. D. Salas, and U. Lall, constitute the subject matter of Chapter 2. In the stochastic simulation of precipi- tation, the writers discuss continuous precipitation models; models of cumulative precipitation over nonoverlapping time intervals, in- cluding Markov chain models, alternating renewal models, and models for precipitation amount; nonparametric models for simu- lating precipitation, including kernel density estimators and kernel- near-neighbor models; and precipitation disaggregation models. Stochastic streamflow simulation comprises continuous time to hourly simulation; weekly, monthly, and seasonal streamflow sim- ulation at a single site; annual streamflow at single site; monthly streamflow simulation; temporal and spatial disaggregation mod- els; nonparametric streamflow simulation models for both single sites and multiple sites; and extensions of the kernel-near-neighbor resampling approach. The treatment is lucid and comprehensive. Written by K. K. Yilmaz, J. Vrugt, H. V. Gupta, and S. Sorooshian, Chapter 3 is devoted to model calibration in watershed hydrology. Beginning with a discussion of approaches to parameter estima- tion for watershed models, including an overview of the manual calibration approach and automated calibration approaches, the chapter goes on to discuss single criterion automated calibra- tion methods; the shuffled complex evolution, the University of Arizona approach; multicriteria calibration methods, including si- multaneous multicriteria calibration, stepwise multicriteria calibra- tion, and the multicriteria constraining approach; automated calibration of spatially distributed watershed models; treatment of parameter uncertainty, including the random walk metropolis algorithm, differential evolution adaptive metropolis, and sequen- tial data assimilation; and parallel computing. It is a well-written chapter providing an up-to-date account of model calibration. The subject matter of Chapter 4 is scaling and fractals in hydrol- ogy, written by D. Veneziano and A. Langusis. Introducing the concepts of fractality and scale invariance, the chapter goes on to discuss fractal and scale-invariant sets, including fractal sets and fractal dimensions and scale-invariant sets and their dimensions; scale invariance for random processes comprising self-similar processes, scalar multifractal processes, and multifractal fields and vector-valued processes; generation of scale-invariant random processes entailing the relationship between scale-invariant and stationary processes, scale-invariant processes as renormaliza- tion limits, and scale invariant processes as weighted sums and products, scale-invariant processes from fractional integration of alpha-stable measures, and processes with limited scale invariance; properties of scale-invariant processes, including H-sssi processes, moment scaling of multifractal processes, existence, moments, and distributions of stationary multifractal measures, and extremes of stationary mulifractal measures and origin of multifractality; fore- casting and downscaling of stationary multifractal measures involv- ing forecasting and downscaling; methods of inference of scaling from data; selected applications in hydrology, with particular focus on rainfall, fluvial topography, floods, and flow through porous me- dia. The treatment in the chapter is comprehensive, up-to-date, well presented, and balanced. Remote sensing for precipitation and hydrologic applications are covered in Chapter 5, written by E. N. Anagnostou. Beginning with a discussion of prediction accuracy and surface modeling, the chapter covers precipitation nowcasting from sparse-based plat- forms; data uses in precipitation forecasting; and data uses in hydrology comprising soil moisture, flood forecasting, and water management; and concludes with an outlook on the future. This is a short but nicely written chapter. Chapter 6, by R. J. Abrahart, M. See, C. W. Dawson, A. Y. Shamseldin, and R. L. Wilby, presents nearly two decades of neural network (NN) hydrologic modeling. Providing a short history of neural networking, the chapter discusses the spread of the field of NN modeling within hydrologic sciences; establishing the field; enlarging the field including traditional regression-type applica- tions, merging of technologies, modular and ensemble modeling, and neuro-hybrid modeling; taking steps to deliver goods, compris- ing building collective intelligence, deciphering of internal compo- nents, and estimating confidence and uncertainty; evaluating the field by asking a series of questions: (1) can NN be made to reveal any physics? (2) Can an optimal training set be identified? (3) Can NN improve on time series analysis? (4) Can NN training be made more adaptive? (5) Are NN good extrapolators? The chapter con- cludes with final thoughts on searching for a killer App. The chap- ter is a well-developed joy to read. Evolutionary computing in hydrology is presented in Chapter 7, written by V. Babovic and R. Rao. Beginning with a discussion of systems and processes, it goes on to discuss evolutionary com- puting comprising evolution principle, evolutionary computing in hydrology; genetic algorithm and its scope in hydrology, entailing modeling the observations, modeling the error, and modeling the model; and issues pertaining to genetic computing, such as selection of data sets, EC setting, model structure, and defect of noise. The chapter is short but to the point. 966 / JOURNAL OF HYDROLOGIC ENGINEERING © ASCE / NOVEMBER 2011 J. Hydrol. Eng., 2011, 16(11): 966-967 Downloaded from ascelibrary.org by 44.192.18.17 on 06/17/22. Copyright ASCE. For personal use only; all rights reserved.