Pairing simple and complex models could improve predictions

Environmental management often relies on complex numerical models. Such models can represent complex natural processes in detail but generally take a lot of computational time to run and even greater computational effort to calibrate. Associating uncertainties with predictions made by these models can also be difficult. Conversely, simple models run faster and are easier to calibrate, but they often leave out key physical details or make simplifying assumptions that could introduce inaccuracy and inhibit uncertainty characterization. To overcome the shortcomings of both types of models, Doherty and Christensen developed a method that uses paired simple and complex models to improve predictive accuracy and quantify predictive uncertainty. The approach could be applied to problems in hydrology, ecology, and many other areas of environmental or climate science. ( Water Resources Research , doi:10.1029/2011WR010763, 2011)