Determination of Uncertainty in Measured Streamflow and Water Quality Data and Application to Model Evaluation

Abstract

In spite of the importance and even scientific responsibility to address uncertainty related to hydrologic and water quality measurement, uncertainty estimates corresponding to measured data are rarely made. The lack of uncertainty estimates can be attributed to the previous lack of a straightforward method to realistically quantify uncertainty. The recent development of fundamental methods to quantify the uncertainty inherent in measured hydrologic and water quality data, however, should increase the application of uncertainty estimates to measured data. If uncertainty estimates are included with measured data sets and adequately communicated to scientists, public interests, and decision makers, then optimal monitoring project design, enhanced model-based decision making, and improved stakeholder understanding will result. The primary objectives of this presentation are: 1) to describe a method for realistic estimation of uncertainty in measured streamflow and water quality data and 2) to illustrate its application in several case studies. The discussion and results presented focus on uncertainty related to discharge measurement, sample collection, sample preservation/storage, and laboratory analysis procedures for measurement of streamflow, nitrogen (N), phosphorus (P), and total suspended sediment (TSS) data from small watersheds. It is hoped that this method (with supporting data and field form templates) will assist monitoring project personnel in making uncertainty estimates for their measured data. The case study results will provide uncertainty estimates associated with individual procedural steps and for the resulting data under a range of monitoring conditions. A secondary objective is to introduce modified goodness-of-fit indicators that consider measurement uncertainty in hydrologic and water quality model evaluation.