Author(s)

Erik D. Kabela^*, Gregory J. Carbone

Department of Geography, University of South Carolina, Columbia, South Carolina, USA.

This paper investigates dynamically downscaled regional climate model (RCM) output from the North American Regional Climate Change Assessment Program (NARCCAP) for two sub-regions of the Southeast United States. A suite of four statistical measures were used to assess model skill and biases were presented in hindcasting daily minimum and maximum temperature and mean precipitation during a historical reference period, 1970-1999. Most models demonstrated high skill for temperature during the historical period. Two outliers included two RCMs run using the Geophysical Fluids Dynamics Lab (GFDL) model as their lateral boundary conditions; these models suffered from a cold maximum temperature bias. Improvement with GFDL-based projections of maximum temperature was noted from May through November when they ran with observed seasurface conditions (GFDL-timeslice), particularly for the east sub-region. Precipitation skill proved mixed-relatively high when measured using a probability density function overlap measurement or the index of agreement, but relatively low when measured with root-mean square error or mean absolute error, because several models overestimated the frequency of extreme precipitation events.

NARCCAP, Model Skill, Model Bias

Kabela, E. and Carbone, G. (2015) NARCCAP Model Skill and Bias for the Southeast United States. American Journal of Climate Change, 4, 94-114. doi: 10.4236/ajcc.2015.41009.