Do Nash values have value?

• DOI: 10.1002/hyp.6825
• Publication Date: 2007-06-18T10:17:50Z
• Authors: Bettina Schaefli; Hoshin V. Gupta
• Keywords: 13. Climate action; 0207 environmental engineering; 02 engineering and technology

How Do We Communicate Model Performance?The process of model performance evaluation is primary importance, not only in the development and calibration process, but also when communicating results to other researchers stakeholders.The basic 'rule' that every modelling result should be put into context, for example, by indicating using appropriate indicators, highlighting potential sources uncertainty, this practice has found its entry large majority papers conference presentations.While question how communicate a end-users currently receiving increasing interest (e.g.Pappenberger Beven, 2006), we-as well as many colleagues-observe regularly take much less care amongst ourselves.We seem assume we are speaking about familiar concepts they have comparable significance various types applications case studies.In doing so, do pay sufficient attention making clear what values represented our measures really mean.Even simple bias between an observed simulated time series need proper context: whereas 10% simulation discharge may unacceptable climate change impact assessment, it concern context real-time flood forecasting.While some can absolute meaning, such common measure linear correlation, vast measures, particular quadratic-error-based properly interpreted viewed reference value.For hydrologists, Nash-Sutcliffe efficiency (Nash Sutcliffe, 1970) (Equation (1)) become part everyday jargon reporting catchment study.Somehow, ignore fact modellers fields environmental sciences often with 'Nash value' is.More important, worth asking whether ourselves know means report value 0•87'.The reality Nash value, while convenient normalized (-inf 1•0) does provide reliable basis comparing different stating this, pointing out anything already known (e.g.Martinec Rango, 1989; Legates McCabe, 1999; Seibert, 2001), think recalling, hope following discussion will encourage hydrologists use more useful manner-which always values-so reported interpreted. Values Need BaselineThe 1970), called, hereafter, (NSE), computed follows: