Meeting:Abstract 2011 CSDMS meeting-075
| You probably paid. If not: | 
|
"USA" is not in the list (Afghanistan, Albania, Algeria, Andorra, Angola, Antigua and Barbuda, Argentina, Armenia, Australia, Austria, ...) of allowed values for the "Country member" property.
Browse abstracts
Mary Hill choose to not submit an abstract for this conference.
[[Image:|300px|right|link=File:]]This work proposes that environmental models would be made more effective by establishing a base set of model fit, sensitivity, and uncertainty measures for which results are reported consistently, along with any other methods considered, and as innovations in model analysis continue. Indeed, there have been many innovations in model analysis in recent decades. While innovation is important, when every subfield within environmental modeling and almost every report use different model analysis methods, the models are difficult to compare and the utility and limitations of the models are more often obscured than elucidated because few are familiar with the analysis method. Adoption of a base set of consistently used methods can be thought of as contributing to the “how” for achieving the model transparency and refutability encouraged by recent essays describing general goals and philosophies for environmental modeling. The phase I set of methods proposed here includes very computationally frugal methods that are most accurate when the model is sufficiently linear. Computationally frugal methods are advantageous when considering models with long execution times typical when considering environmental systems. Recent work indicates that many of the nonlinearities that have caused so much trouble in the past are numerical artifacts without validity to real world system dynamics. It appears that robust models absolved of such artifacts can often be fruitfully evaluated using linear methods. Second, situations in which the computationally frugal methods fail are generally apparent rather quickly, so little time is lost relative to the effort required by computationally demanding methods. Comparisons of linear methods to selected computationally expensive methods demonstrate the relative utility of both. Establishing base sets of linear (Phase I) and nonlinear (Phase II) methods provides a foundation that is expected to increase utility of environmental model results and interest in sensitivity analysis and uncertainty methods. The increased utility and interest would benefit all developers and users of environmental models, including policy makers and resource managers.
