Statement on Measurement and Model Uncertainty
Daren Harmel, October 2013
About 10 years ago, based on recent work by many scientists, most notably Keith Beven, Richard Cooper, and Ken Reckhow, I came to realize the importance of estimating uncertainty in hydrologic and water quality monitoring projects and communicating that uncertainty to users of the measured data and model results. Beven (2006) and Pappenberger and Beven (2006) eloquently yet firmly state that every hydrologic and water quality monitoring and research project should estimate and discuss the uncertainty in the results. Because of the numerous benefits of uncertainty analysis and communication, this statement is difficult to argue with. Support of this statement is certainly not a condemnation of previous scientific efforts that have ignored uncertainty, as regretfully I have also been guilty of this oversight in the past. Instead, it is meant to encourage scientists to evaluate uncertainty in future projects. The measured data and model predictions we produce provide the basis for many decisions with important environmental and socio-economic ramifications; thus, the uncertainty in those data should no longer be ignored.
Uncertainty in Runoff and Sediment Data
Measured at the USDA-ARS Riesel Watersheds
Much of the Riesel data was formatted, quality checked, and placed on the web by Kevin King and I in the late 1990’s and early 2000’s. This work occurred prior to my realization of the importance of measurement uncertainty; therefore, the Riesel data are presented without corresponding uncertainty estimates. I have no plans at the current time to revisit the historical Riesel data and make uncertainty estimates for several reasons: 1) the data set is extremely large (e.g. runoff ~1300 site years, rainfall ~1400 site years), and 2) the methods used and associated “meta data” comments on data errors are not readily available. However, the uncertainty estimates presented below, which are based on recent work at Riesel (Harmel et al., 2009), should reasonably apply to the historical runoff and sediment data from the Riesel watersheds.
Beven, K. 2006. On undermining the science? Hydrol. Process. 20:3141-3146.
Harmel, R.D., D.R. Smith, K.W. King, and R.M. Slade. 2009. Estimating storm discharge and water quality data uncertainty: A software tool for monitoring and modeling applications. Environ. Modelling Software 24: 832-842.
Pappenberger F., and K.J. Beven. 2006. Ignorance is bliss: 7 reasons not to use uncertainty analysis. Water Resources Res. 42(5), doi:10.1029/2005W05302