Reconstruction of hydrologic and climatic variability in the Colorado River Basin

This dissertation investigated the reconstruction of hydrologic and climate variability in the Colorado River Basin for the past 500 years. Unimpaired streamflow stations and regional April 1 snow water equivalent of the basin were reconstructed using tree-ring chronologies as predictors in Partial Least Squares Regression (PLSR). Regional April 1 snow water equivalent time series was developed with principal component analysis and cluster analysis on snow course stations in the basin. For the reconstruction of unimpaired streamflow and regional April 1 snow water equivalent, all available standard tree-ring chronologies inside the Colorado River Basin were screened based on the correlation criterion (\u3e95% significance level). Then, PLSR was run using the cross validation approach (i.e. removing the least correlated screened tree-ring chronologies) to obtain an optimal result. Further, a PLSR reconstruction was compared with the different reconstruction procedures (Stepwise Linear Regression and Stepwise Principal Component Regression) in order to find the preferred method for climate variable reconstruction. Finally, the non-parametric rank sum test was performed to evaluate the individual and coupled impact of interannual and interdecadal ocean climate phenomenon on climate variables (streamflow and snowpack) of the basin using extended records; The major contributions of this research are divided into three major categories. First, unimpaired streamflow stations in the Colorado River basin were reconstructed using the Partial Least Square Regression (PLSR) technique. The current practice of reconstructing streamflows by using multiple linear regression and principal component regression were compared with the performance of Partial Least Square Regression technique (PLSR) based on cross validation standard error. The spatial and temporal variability of drought was evaluated for all the unimpaired streamflow stations and the different centuries in the record. Second, snowpack was regionalized using principal component and cluster analysis which helps to identify coherent regions in the basin. Further, regionalized snowpack was reconstructed for the past 500 years using the PLSR in order to understand the long-term regional spatial and temporal variability of drought in the basin. Third, the teleconnection study between interannual and interdecadal ocean climate phenomenon and climate variables (streamflow and snowpack) in the region was performed using reconstructed data considering different lag years (0, +1, +2 and +3); The outcome of this research identified an improved reconstruction approach to extend the climate variables (unimpaired streamflow and snowpack), regionalized climate variables, and determined the relationship between reconstructed climate variables and reconstructed large-scale ocean atmospheric patterns. These outcomes will help to improve the current practice of drought management planning and drought forecasting in the western United States