Land surface temperature （LST） is a very important parameter controlling the energy and water balance between the atmosphere and the land surface. But when the sky is overcast

LST measurements are impossible with thermal remote sensing. Consequently

only cloudy-free measurements are useful

implying that the data set will be biased. Choosing the best spatial interpolation of land surface-the gradient plus inverse distance squared （GIDS） method

estimating LST under cloud cover is possible. An evaluation on LST estimation under cloud cover using GIDS in Jiangning study site of China Jiangsu province was tested in this paper with ETM+ and ASTER GDEMV1 data. The results showed that the GIDS can estimate LST smoothly with high accuracy. The error increased with cloud cover extent expanding

and the maximum of mean absolute error（MAE）

root mean squared error （RMSE） were less than 0.9℃ and 1.2℃

respectively. When the cloud cover extent was less than 100×100 pixel

the maximum error of MAE and RMSE were lower than 0.8℃ and 1℃

respectively. The accuracy of GIDS varied with the unclouded typical pixel

complexity of spatial constructer and land cover type of cloud cover. An important parameter

the standard deviation STD of Normalized Difference Vegetation Index（NDVI） for overcast

can index the uncertainty of GIDS estimated results and have same trend with MAE and RMSE. It can detect the estimated error extent and evaluate the estimated result linking to the denotation of STD of NDVI for spatial complexity and NDVI-LST negative correlation in cloudy condition. Therefore

care should be given to the GIDS estimated result before the application.