A framework for quantifying and incorporating climate data uncertainty into water resources assessment

Abstract

Rainfall and temperature (the main driver of evaporation) are key inputs for hydrologic models in studying catchment responses to climate scenarios. Both rainfall and temperature, however, are uncertain, with rainfall having a larger degree of uncertainty. Using uncertain inputs in hydrologic models, without due consideration of their associated uncertainties, results in biased outcomes. The purpose of this thesis is to develop methods for quantifying uncertainties in climate data (with emphasis on rainfall) towards proposing strategies to incorporate these uncertainties into water resource assessment. Rain gauge and satellite rainfall data are initially compared and merged to produce an improved gridded rainfall dataset with its associated standard error. This is implemented for Australian rainfall. The standard error estimation logic is then extended to develop a novel uncertainty metric, the square root error variance (SREV), for quantifying uncertainties in global climate model (GCM) data. The method is applied to estimate GCM-projected rainfall and temperature uncertainty across the world. It is found that GCM uncertainty arises mainly from model structural errors. Subsequently, two case studies that implement the SREV metric into hydrologic systems are carried out. First, future drought, across the world, is estimated with due consideration to the uncertainties involved in GCM rainfall projections. Simulation extrapolation, which reduces parameter bias when input errors are known, is used to mitigate biases in drought estimates. It is found that consideration of GCM rainfall uncertainties is vital, as drought values with and without considering the uncertainties are significantly different. Second, a comprehensive analysis is carried out to evaluate water availability at the Warragamba Catchment in Sydney, Australia. An additive error model is proposed to generate rainfall and temperature realizations that are used to simulate streamflow. Future storage requirement with its associated uncertainty is then evaluated using reservoir behavior analysis. It is found that the existing storage capacity suffices the future requirements, although large uncertainty exists in storage estimates. In conclusion, the thesis presents methods to quantify and account for uncertainties in key hydrologic variables. Provision of these uncertainties offers an effective platform for risk-based assessments of any integrative or adaptive water management plans that may be formulated using measured or simulated climate data.