Date of Completion

12-18-2011

Embargo Period

12-23-2011

Advisors

John Silander, Richard Anyah

Field of Study

Environmental Engineering

Degree

Master of Science

Open Access

Open Access

Abstract

Global Climate Models (GCMs) are the typical sources of future climate data required for impact assessments of climate change. However, GCM outputs are related to model-related uncertainties and involve a great deal of biases. Bias correction of model outputs is, therefore, necessary before their use in impact studies. The coarse resolution of GCM simulations is another hindrance to their direct use in fine-scale impact analysis of climate change. Although downscaling of GCM outputs can be performed by dynamical downscaling using Regional Climate Models (RCMs), it requires large computational capacity. When daily climate data from multiple GCMs are required to be downscaled, dynamical downscaling may not be a feasible option. Statistical downscaling, in contrast, can be efficiently used to downscale a large number of GCM outputs at a fine temporal and spatial scale. This study performs the bias correction and statistical downscaling of daily maximum and minimum temperature and daily precipitation data from six GCM and four RCM simulations for the northeast United Stated (US). The spatial resolution of the data set is 1/8°x 1/8° and it spans from 2046 to 2065. This fine-scale daily climate data set, which has been created using Bias Correction and Spatial Downscaling (BCSD) approach, can be directly used in regional impact studies for the northeast US.

Using both raw and bias corrected daily precipitation data from two GCMs and two RCMs, one extreme precipitation index has been analyzed for the observed climate. The comparison between the results demonstrates that bias correction is important not only for GCM outputs, but also for RCM outputs. When the same analysis has been performed for future climate, bias correction has led to a larger level of agreements among the models in predicting the magnitude and capturing the spatial trend for the extreme precipitation index. Moreover, five extreme climate indices have been analyzed at 1/8° spatial resolution for future climate using the bias corrected and statically downscaled data from multiple GCMs and RCMs. The incorporation of dynamical downscaling as an intermediate step has not led to any considerable changes from the results of statistical downscaling. Statistical downscaling with bias correction has been sufficient to create a fine-scale daily climate data set to be directly used in impact studies. The future means of five extreme climate indices, which have been calculated from GCM and RCM ensembles, have been compared to their observed means. The decrease in total number of frost days because of the future warming will be similar over the entire northeast region. The earlier arrival of spring will lead to an extended growing season and the magnitude of the changes will be larger in the coastal area. The comparison of precipitation extreme indices indicates an increase in the heavy precipitation events in future climate for most of the region.

Major Advisor

Guiling Wang

Share

COinS