Landsat-8 data (level 1T) received by users are still in digital number and can be used directly for mapping land use/land cover. However, the data still has low radiometric accuracy when it is used to derive information such as vegetation index, biomass, land use/land cover classification, etc. so that so that it requires radiometric/atmospheric correction. In this study, we use atmospheric correction method of the second simulation of satellite in the solar spectrum (6S) and Fast Line-of-sight Atmospheric Analysis of Spectral Hypercube (FLAASH) to eliminate atmospheric influences and compare the results with field measurements. The atmospheric parameters used were aerosol optical depth (AOD), water vapour column and ozone thickness from MODIS data with the date and time of acquisition approaching with Landsat-8 data. From the analysis conducted on the spectral response of atmospheric corrected image shows the 6S model has better accuracy for the spectral response from the rice growth phase compared to the FLAASH model. The analysis of the values of vegetation indices (NDVI, EVI, SAVI and MSAVI) shows that the 6S model has better accuracy for NDVI while for EVI, SAVI and MSAVI the FLAASH model has slightly better accuracy than 6S.