A nonlinear statistical downscaling approach of ECHAM 5 model data to project heavy precipitation events for Oahu, Hawaii

Abstract

The objective of this thesis is to build, calibrate and test nonlinear statistical models in an attempt to find the optimal relationship between large-scale atmospheric variables provided by coarse resolution global models and station specific heavy precipitation data on the island of Oahu, Hawaiʻi. The models will be calibrated using NCEP reanalysis II data and tested with an independent data set for model verification. After the models are adequately calibrated and tested, GCM data are used as input into the calibrated statistical models for the period 2011-2040. A BCa bootstrap resampling method is used to provide 95% confidence intervals of the storm frequency and intensity for all three datasets (actual observations, downscaled GCM output from the present-day climate, and downscaled GCM output for future climate). This provides a method to analyze future heavy precipitation at the station scale and can provide the needed information to best plan and prepare for changes in heavy precipitation events. Results suggest a tendency for increased frequency of heavy rainfall events, but a decrease in rainfall intensity during the next thirty years (2011-2040) for the southern shoreline of Oahu.