Electronic Thesis and Dissertation Repository

Degree

Doctor of Philosophy

Program

Civil and Environmental Engineering

Supervisor

Slobodan P. Simonovic

Abstract

Assessment of climate change impacts on hydrology at watershed scale incorporates (a) downscaling of global scale climatic variables into local scale hydrologic variables and (b) assessment of future hydrologic extremes. Atmosphere-Ocean Global Climate Models (AOGCM) are designed to simulate time series of future climate responses accounting for human induced green house gas emissions. The present study addresses the following limitations of climate change impact research: (i) limited availability of observed historical information; (ii) limited research on the detection of changes in hydrologic extremes; and (iii) coarse spatio-temporal resolution of AOGCMs for use at regional or local scale. Downscaled output from a single AOGCM with a single emission scenario represents only a single trajectory of all possible future climate realizations and cannot be representative of the full extent of climate change. Present research, therefore addresses the following questions: (i) how should the AOGCM outputs be selected to assess the severity of extreme climate events?; (ii) should climate research adopt equal weights from AOGCM outputs to generate future climate?; and (iii) what is the probability of the future extreme events to be more severe? Assessment of regional reanalysis hydro-climatic data has shown promising potential as an addition to the observed data in data scarce regions. A new approach using statistical downscaling based nonparametric data-driven kernel estimator is developed for quantifying uncertainties from multiple AOGCMs and emission scenarios. The results are compared with a Bayesian reliability ensemble average method. The generated future climate scenarios represent the nature and progression of uncertainties from several global climate models and their emission scenarios. Treating the extreme precipitation indices as independent realization at every time step, the kernel estimator provides variable weights to the multi-model quantification of uncertainties. The probabilities of the extreme indices have added useful insight into future climate conditions. Finally, the current method of developing future rainfall intensity-duration-frequency curves is extended by introducing a probabilistic weighted curve to include AOGCM and emission scenario uncertainties using the plug-in kernel. Present research has thus expanded the existing knowledge of dealing with the uncertainties of extreme events.

Share

COinS