UC Santa Barbara

Water resources managers must increasingly consider climate change implications of, whether the concern is floods, droughts, reservoir management, or reliably supplying consumers. Hydrologic and water quality modeling of future climate scenarios requires understanding global climate models (GCMs), emission scenarios and downscaling GCM output, since GCMs generate climate predictions at a resolution too coarse for watershed modeling. Here we present theoretical considerations needed to understand the various downscaling methods. Since most watershed modelers will not be performing independent downscaling, given the resource and time requirements needed, we also present a practical workflow for selecting downscaled datasets. Even given the availability of a number of downscaled datasets, a number of decisions are needed regarding downscaling approach (statistical vs. dynamic), GCMs to consider, options, climate statistics to consider for the selection of model(s) that best predict the historical period, and the relative importance of different climate statistics. Available dynamically-downscaled datasets are more limited in GCMs and time periods considered, but the watershed modeler should consider the approach that best matches the historical observations. We critically assess the existing downscaling approaches and then provide practical considerations (which scenarios and GCMs have been downscaled? What are some of the limitations of these databases? What are the steps to selecting a downscaling approach?) Many of these practical questions have not been addressed in previous reviews. While there is no “best approach” that will work for every watershed, having a systematic approach for selecting the multiple options can serve to make an informed and supportable decision.