Geolocation and Processing
The scenes used to create the image maps were selected from a combination of Collections 4, 5, and 6 MODIS/Aqua and MODIS/Terra data sets (MYD02QKM and MOD02QKM). Band 1 and Band 2 scenes from MYD02QKM and MOD02QKM, together with illumination and viewing angles from MYD03 and MOD03¹, were geolocated and resampled onto the projection grid using NSIDC's MODIS Swath-to-Grid Toolbox (MS2GT). The software interpolated the MYD03 and MOD03 latitude/longitude data from 1 km resolution to 250 m and then resampled the MODIS/Aqua and MODIS/Terra Level-1B calibrated radiances to the grid using a forward elliptical weighted average algorithm (Greene et al., 1986).
¹MODIS Level 1A Geolocation Fields from EOS Aqua and EOS Terra
Destriping of MODIS Image Data
The MS2GT algorithm was modified to remove MYD02QKM and MOD02QKM striping artifacts, a known problem with all Terra and Aqua MODIS 250 m Level-1B data, by adding a Lambertian solar zenith angle normalization on the swath data for both bands. Telemetry noise and line drops, which have the appearance of chads in the projected images, were reset to zero (treated as masked cloud areas). This procedure is discussed in detail in the Destriping of MODIS Image Data section of the MOA2004 documentation.
Cloud Masking
The geolocated scenes were manually masked to remove clouds, cloud shadows, fog, blowing snow, and heavy surface frost. Refer to the Cloud Masking section of the MOA 2004 documentation. The final image maps are nearly perfectly cloud-cleared, except for some areas of thin clouds, cirrus cloud shadows, and fog or low-lying small clouds.
Surface Morphology Image Map
Geolocated and destriped Band 1 images were high-pass filtered to reduce non-Lambertian illumination and to reset the mean grayscale range to a common value for compositing. For each gridded image, the investigators created a corresponding weight image in which each non-masked pixel is assigned a scalar value or weight. Weights were computed based on proximity to the nadir track, favoring near-nadir areas, and proximity to an image or mask edge to feather the edges of the component images. Finally, weight images were then combined using stacking techniques called image super-resolution or data cumulation. These techniques allow multiple images to contribute to how a single grid cell is represented in the final composite.
The algorithms used to compute and combine the weight images into the final mosaics are provided in the MOA2004 Compositing the Image Swaths documentation.
Error Sources
Wolfe et al. (2002) estimated the accuracy of the MYD03/MOD03 Level-1A geolocation data to be 50 m, considerably better than the MYD02QKM/MOD02QKM ground-equivalent nadir pixel size of 250 m. The accuracy and precision of this geolocation was also tested for the MOA2004 using known surface sites, such as South Pole Station, Vostok Station, Siple Dome camp and traverse trail, and areas of well-mapped coastline such as Ross Island and the northern Antarctic Peninsula. The investigators did not find discrepancies greater than 125 m in the projected location of any fixed object.