Developing algorithm for operational GOES-R land surface temperature product

tellite (GOES) program is developing the Advanced Baseline Imager (ABI), a new generation sensor to be carried onboard the GEOS-R satellite (launch expected in 2014). Compared to the current GOES Imager, ABI will have significant advantages for re-trieving land surface temperature (LST) as well as providing qual-itative and quantitative data for a wide range of applications. The infrared bands of the ABI sensor are designed to achieve a spatial resolution of 2 km at nadir and a noise equivalent temperature of 0.1 K. These improve the imager specifications and compare well with those of polar-orbiting sensors (e.g., Advanced Very High Resolution Radiometer and Moderate Resolution Imaging Spec-troradiometer). In this paper, we discuss the development of a split window LST algorithm for the ABI sensor. First, we simulated ABI sensor data using the MODTRAN radiative transfer model and NOAA88 atmospheric profiles. To model land conditions, we developed emissivity data for 78 virtual surface types using the surface emissivity library from Snyder et al. Using the simulation results, we performed regression analyses with the candidate LST algorithms. Algorithm coefficients were stratified for dry and moist atmospheres as well as for daytime and nighttime conditions. We estimated the accuracy and sensitivity of each algorithm for different sun-view geometries, emissivity errors, and atmospheric assessments. Finally, we evaluated the most promising algorithm using real data from the GOES-8 Imager and SURFace RADiation Network. The results indicate that the optimized LST algorithm meets the required accuracy (2.3 K) of the GOES-R mission