Inter-satellite atmospheric and radiometric correction for the retrieval of Landsat sea surface temperature by using Terra MODIS data

Abstract

Thermal infrared images of Landsat-5 TM and Landsat-7 ETM+ sensors have been unrivalled sources of high resolution thermal remote sensing (120 m for TM and 60 m for ETM+) for more than two decades. As the sensors have only one thermal channel, however, the correction of atmospheric effect has been virtually limited, degrading the accuracy of sea surface temperature (SST) measurement. Launched in 1999, MODIS sensor onboard Terra satellite is equipped with two thermal channels that can provide accurate atmospheric correction at 1 km resolution. In this paper we propose an inter-satellite calibration method to correct the radiometric and atmospheric effect of Landsat brightness temperature by using the atmospherically corrected Terra MODIS SST which lags Landsat pass by 30 minutes only. Comparison of the corrected Landsat SST with in situ SST near the coast of South Korea showed a significant improvement in root mean square error from 2.31 °C before the correction to 0.96 °C after the correction. Errors from spatial and temporal inhomogeneities over 1 km × 1 km window could be masked out by identifying negative correction term and applying a root mean square deviation criterion between Landsat and MODIS SSTs. We expect that Landsat SST product obtained after the launch of Terra can be atmospherically corrected by using the method proposed in this paper while maintaining the merit of high-resolution Landsat thermal infrared imagery.