Analysis and Improvement of Geo-Referencing Accuracy in Long Term Global AVHRR Data

Precise geolocation is one of the fundamental requirements for generating high-quality Advanced Very High Resolution Radiometer (AVHRR) Satellite Climate Data Record (SCDR) at 1-km spatial resolution for climate applications. The Global Climate Observing System (GCOS) and Committee on Earth Observing Satellites (CEOS) identified the requirement for the accuracy of geolocation of satellite data for climate applications as 1/3 field-of-view (FOV). This requirement for AVHRR series on the National Oceanic and Atmospheric Administration (NOAA) platforms cannot be met without implementing the ground control point (GCP) correction, especially for historical data, because of the limited accuracy of orbit models and uncertainty in the satellite attitude angles. This work presents a new analysis of the geo-referencing accuracy of global AVHRR data, that uses an automated image matching at pre-selected GCP locations. As a reference image, we have been using the clear-sky monthly composite imagery derived from Moderate Resolution Imaging Spectroradiometer (MODIS) MOD09 dataset at 250-m resolution. The image matching technique is applicable to processing not only the daytime observations from optical solar bands, but also the nighttime imagery by using the long wave thermal channels. The method includes the ortho-rectification to correct for surface elevation and achieves the sub-pixel accuracy in both along-scan and along-track directions. The produced image displacement map is then used to derive a correction to satellite clock error and the attitude angles. The statistics and pattern of these corrections have been analyzed for different NOAA Polar-orbiting satellites by using the HRPT, LAC, and GAC data sets. The application of the developed processing system showed that the algorithm achieved better than 1/3 FOV geolocation accuracy for most of AVHRR 1-km scenes. It has a high efficiency rate (over 97%) for global AVHRR data from NOAA-6 through NOAA-19.