Precipitation and runoff calculated by climate models suffer various deficiencies because of their low spatial resolution and incomplete process treatment. Thus, efforts have continued to compensate for these shortcomings in the use of model output through inclusion of additional process parameterizations.
The use of model output to determine river flows has been addressed by Miller and Russell [1992] and Liston et al. [1994]. Gao and Sorooshian [1994] have developed a statistical procedure using observational data to transform climate model precipitation into a realistic frequency and intensity distribution.
Robock et al. [1993] discuss the general use of GCM simulations, with their common errors, for use in the study of required impacts of future climate change. They suggest a new procedure for the combination of GCM output with climate information in order to produce future scenarios, even when the GCMs give poor regional simulations.