Review of Statistical Downscaling for Hydrological and Environmental Applications by Taesam Lee and Vijay P. Singh

The term downscaling is commonly used to represent the myriad ways that address the mismatch between the coarse temporal and spatial scales employed in global circulation models (GCMs) and finer resolution projections that are needed for hydrologic and environmental applications. Dynamical downscaling embeds physical models [usually regional climate models (RCMs)] to successively obtain variables at finer spatial and temporal resolutions using coarse-resolution GCM information as boundary conditions, while statistical downscaling seeks to achieve the transfer to finer scales typically through regression, weather generation, or weather typing approaches. The latter is the subject matter of this book by Lee and Singh. Given the recent propensity for statistical downscaling in the scientific literature, the authors do the community a service by summarizing recent advances in one convenient location.

The book has been laid out in nine chapters. The introductory chapter provides the framework for downscaling and describes the strengths and limitations of both dynamical and statistical downscaling methods, while the second chapter provides a quick review of basic concepts in statistics (random variables, distributions, parameter estimation, Monte Carlo simulations) to prepare users for later chapters. The third chapter describes how to access GCM, RCM, and reanalysis data sets.

The fourth chapter familiarizes readers with the idea of bias correction—a necessary operational step because of scale mismatch between model resolutions and measurements. Popular bias correction methods such as adjustment techniques and quantile mapping are discussed here.

The next three chapters take us to the heart of statistical downscaling methods. Chapter 5 deals with linear, nonlinear, and stepwise regression techniques. Transfer functions fall into this category. The authors briefly describe artificial neural networks in this chapter. Local hydrologic variables do not always exhibit strong correlations to GCM outputs, thus limiting regression techniques. Weather generator models employing stochastic simulations offer an alternative and are described in Chapter 6. Chapter 7 discusses weather typing models that use classification techniques to condition stochastic weather simulations. Empirical and objective weather typing are discussed in this regard.

Temporal and spatial downscaling are the foci of Chapters 8 and 9, respectively. The popular $K$-nearest neighbor resampling method for temporal downscaling is described, while interpolation and bias correction techniques are the subject of the ninth chapter, on spatial downscaling.

The easy writing style allows novices to grasp the whys and hows of statistical downscaling, while retaining enough rigor to engage advanced researchers as well. The book offers a good compendium of statistical downscaling techniques, and is sure to serve as a baseline for future techniques in this rapidly growing field of hydroclimatology. The book would be an important addition to both personal and public libraries.