Directed: Characterizing the Uncertainty of Hydrologic Variables Generated Using Nonparametric Techniques
Contact: Subhrendu Gangopadhyay, firstname.lastname@example.org 303-445-2465
Fiscal Year: 2011 - present
- How can we characterize uncertainty associated with non-parametric statistical analyses relevant to floods and droughts?
Several nonparametric methods are gaining popularity in their use alongside traditional parametric approaches for stochastic simulation of hydrologic variables. However, these stochastic simulations, and the simulation statistics using the nonparametric methods are strongly impacted by two factors: (1) sample size of the historical data, and (2) number of simulations.
Theory is well established in traditional parametric statistics to quantify the role of sample and simulation sizes on hydrologic simulation uncertainty. But, there has not been much research in the data-driven nonparametric realm. This research will evaluate the power and uncertainty associated with sample size and number of simulations in nonparametric streamflow simulations. Understanding and quantifying this uncertainty is important for realistic estimation of risk for water resources planning and management.
Need and Benefit:There is considerable scientific uncertainty regarding the precise nature of how a changing climate may affect Reclamation’s river basins, and thus water and power operations. This research will help evaluate this uncertainty and provide for more accurate streamflow forecasts and simulations for Reclamation decisionmaking.
Partners: University of Colorado, Boulder
Keywords: statistical analysis, non-parametric approaches, climate change, streamflow forecasts