| Regional regression models can be employed to estimate hydrologic statistics at ungauged river sites. The increased amount of GIS-based data describing watershed characteristics has created situations with a large number of highly correlated watershed characteristics. The use of ordinary least squares (OLS) regression procedures with highly correlated variables can produce multicollinearity. This thesis developed a Monte Carlo simulation to compare four techniques for handling multicollinearity: OLS, OLS with Variance Inflation Factor (VIF) screening, principal component regression (PCR), and partial least squares regression (PLS). The impact of multicollinearity is magnified at smaller samples sizes, higher correlations, and larger model error variances. If one is only interested in model predictions within the data range, the use of OLS appears warranted as the complexity of using biased regression techniques to address multicollinearity does little to improve model predictions. A case study from the eastern United States also indicated that OLS generally provides better or similar prediction estimators as VIF, PCR, and PLS. |
