What has been accomplished twenty years after the Oreskes et al. (1994) critique? Current state and future perspectives of environmental modeling in the Great Lakes

摘要

With well over 1000 citations, the Oreskes et al. (1994) paper stands out as one of the classical critiques of the veracity of the scientific methodology of models in earth sciences, arguing that the validation of models that deal with natural systems is inherently impossible. Going beyond the controversy of the "technical versus philosophical" meaning of validation, this viewpoint reflects the important notion that model outputs should be viewed through the prism of the underlying assumptions and that good model performance in one or more settings is not evidence for general applicability, but rather the start of a perpetual race for predictive confirmation. While the Oreskes et al.'s (1994) critique has been a defining moment of the broader appreciation of the challenges surrounding a model validation exercise the documented inadequacy of many models to address important societal issues reflects the fact that the field has advanced without the healthy dose of introspection required to obtain good science. An evidence of the latter assertion is the inconsistency that still characterizes the environmental modeling practice with respect to the methodological steps typically followed (Arhonditsis and Brett, 2004; Arhonditsis et al., 2006; Stow et al., 2009; Robson, 2014; Wellen et al., submitted for publication). After more than four decades of active modeling in the context of environmental management and policy analysis, many of the published aquatic ecosystem and watershed modeling studies still fail to report the results of predictive confirmation, goodness of fit statistics, and uncertainty analysis in the broader sense (Fig. 1a).