Atmospheric correction is inevitable in estimating land surface temperature and land surface emissivity using hyperspectral thermal infrared data.The Autonomous Atmospheric Compensation（AAC） and In-scene Atmospheric Compensation（ISAC） are the two main methods for atmospheric correction of hyper-spectral thermal data.Those two methods were applied to the simulated datasets.Results show that the high accuracy of the AAC method

except for the tropic atmosphere

and the root mean stand error（RMSE） of the transmittance and the up-welling radiance for the AAC method is less than 0.002 and 0.004 W.m-2.sr-1.cm

respectively.On the contrary

the error of the ISAC method is large

and the error of the transmittance varies from 0.05 to 0.3 and the error of the up-welling radiance changes from 0.003 W.m.sr.cm to 0.035 W.m.sr.cm.The error increases as the increase of total precipitable water of the atmosphere.Analysis on the impact of heterogeneity of atmosphere on accuracy of atmospheric correction shows that the accuracy of AAC approach is influenced significantly by the heterogeneity of atmosphere.It is necessary to develop atmospheric correction methods for hyper-spectral thermal infrared data of low spatial resolution to overcome the drawback of requiring homogeneous atmosphere by existing atmospheric correction approaches.