Date of Award




Document Type

Master's Thesis

Degree Name

Master of Science (MS)


Department of Atmospheric and Environmental Sciences

Content Description

1 online resource (ii, v, 55 pages) : illustrations (some color), color maps.

Dissertation/Thesis Chair


Committee Members



GFDL AM4, temperature bias, Atmospheric models, Atmospheric temperature, Atmospheric circulation

Subject Categories

Atmospheric Sciences


The Atmospheric Model version 4.0 (AM4) is the atmospheric component of the latest climate and Earth system models (CM4 and ESM4) developed by NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL) in Princeton, New Jersey. The AM4 makes some improvements in its radiative flux simulations. As a new model, its performance needs to be evaluated. We analyzed the model’s historical simulations (ten-year simulations from 2001 to 2010), in order to evaluate the diurnal cycle of surface air temperature (Tas) simulated by the AM4 model and perform some diagnostic analyses of the surface energy fluxes to help understand the temperature biases. In comparison with the observations and reanalyses, the model Tas has a smaller diurnal range, which is partially contributed by the higher reference height in the model. Other factors that have contributed substantially to the Tas biases include surface sensible heat (SH), latent heat (LH), longwave radiation (LW) and shortwave radiation (SW) fluxes. Each flux contributes differently over different regions. In this study, we paid more attention to four regions: (1) North Africa (10o-30 o N and 15 o W-35 o E); (2) Northeast Asia (55 o -70 o N and 90 o -120 o E); (3) India (9 o -23 o N and 68 o -89 o E); and (4) the Amazon basin (0 o -13 o S and 53 o -70 o W). The largest warm temperature biases occur over Northeastern Asia, which could be related to the vegetation settings in the model (Zhao et al. 2018a). The overestimate in vegetation in the model leads to less reflection of shortwave radiation during daytime, contributing to the warm biases in the daily maximum and mean surface air temperature. Also, more vegetation results in less downward sensible heat transfer at night (Kawashima et al. 2000), contributing to the warm biases in the minimum surface air temperature. We also analyzed surface wind speed, the temperature gradient between surface skin temperature and Tas, and relative humidity to evaluate the possible causes to the biases in surface heat fluxes. Most underestimates in surface heat fluxes are related to weaker wind speed in the model.