A coupled atmosphere-ocean model developed at the Institute for Space Studies at NASA Goddard Space Flight Center (Russell et al., 1995) was used to verify the validity of Haney-type surface thermal boundary condition, which linearly connects net downward surface heat flux Q to air / sea temperature difference △T by a relaxation coefficient κ The model was initiated from the National Centers for Environmental Prediction (NCEP) atmospheric observations for 1 December 1977, and from the National Ocean Data Center (NODC) global climatological mean December temperature and salinity fields at 1°× 1°resolution. The time step is 7.5 minutes. We integrated the model for 450 days and obtained a complete model-generated global data set of daily mean downward net surface flux Q, surface air temperature TA, and sea surface temperature To. Then,we calculated the cross-correlation coefficients (CCC) between Q and △T. The ensemble mean CCC fields show (a) no correlation between Q and △T in the equatorial re- gions, and (b) evident correlation (CCC≥0.7) between Q and △T in the middle and high latitudes. Additionally, we did the variance analysis and found that when κ= 120 W m-2K-1, the two standard deviations, σQ and σκ△T, are quite close in the middle and high latitudes. These results agree quite well with a previous research (Chu et al., 1998) on analyzing the NCEP re-analyzed surface data, except that a smaller value of κ(80 Wm-2K-1) was found in the previous study. A coupled atmosphere-ocean model developed at the Institute for Space Studies at NASA Goddard Space Flight Center (Russell et al., 1995) was used to verify the validity of Haney-type surface thermal boundary condition, which linearly connects net downward surface heat flux Q to air / sea temperature difference ΔT by a relaxation coefficient κ The model was initiated from the National Centers for Environmental Prediction (NCEP) atmospheric observations for 1 December 1977, and from the National Ocean Data Center (NODC) global climatological mean December temperature and the salinity fields at 1 ° × 1 ° resolution. The time step is 7.5 minutes. We integrated the model for 450 days and obtained a complete model-generated global data set of daily mean downward net surface flux Q, surface air temperature TA, and sea ​​surface temperature To. Then, we calculated the cross-correlation coefficients (CCC) between Q and △ T. The ensemble mean CCC fields show (a) no correlation between Q and △ T in (CCC ≧ 0.7) between Q and ΔT in the middle and high latitudes. Additionally, we did the variance analysis and found that when κ = 120 W m-2K-1, These results agree quite well with a previous research (Chu et al., 1998) on analyzing the NCEP re-analyzed surface data, except that the two standard deviations, σQ and σκΔT, are quite close in the middle and high latitudes. a smaller value of κ (80 Wm-2K-1) was found in the previous study.