Uncertainty in hydrological estimation

2021 Spring.Includes bibliographical references.Detailed hydrometeorologic analyses and uncertainty assessments are needed to aid water resources decision-making, to account for upstream-downstream linkages and dominant process scale for integrated land and water resources management and planning. The water balance is a fundamental concept in hydrology that inspires many tools for predicting the specific components including precipitation, streamflow, soil moisture, and groundwater storage. A water balance is typically expressed as an equation that relates water inputs, outputs, and storage of a system. The water balance model is applied to analyze the allocation of water among components of the hydrologic system. Knowledge on the components composing inputs and outputs in a water balance are essential to understanding watershed processes. While methods to measure and model water balance components continue to improve, all components of the balance have substantial uncertainty. Methods to analyze a water balance should acknowledge these uncertainties and consider how they propagate through water balance calculations in order to better assist water resources decisions. This research investigated four water balance components: (1) snowpack sublimation, (2) precipitation as snow, (3) precipitation as rain, and (4) stream discharge in mountainous watersheds in order to examine and build our knowledge of uncertainty in the water balance for mountainous environments. The research presented in this dissertation supports a theme that hydrology is a highly uncertain science, where uncertainty is a result of the hydrologic community's knowledge gap to accurately model physics of atmospheric and hydrologic processes. A finding of this work is that no component of the water balance can be quantified at watershed scale without estimating he associated uncertainty. Results highlight that mean cumulative snowpack sublimation uncertainty is 41% with individual input variable uncertainties in the range of 1 to 29%; simulated to observed basin mean snow depth was estimated within 15% for 10-years while extreme dry and wet years were within 5%; and forcing precipitation datasets used in hydrologic models to estimate streamflow have cumulative uncertainties in the range of 30 to 60%. Results of this dissertation identify the importance to account for uncertainty in water resources, i.e., Monte Carlo methods, to properly account for and quantify associated risks in water management and design infrastructure decisions