A first horizontal accuracy assessment of ICESat-2

Abstract:

This paper assesses the horizontal accuracy of laser points derived from observations from NASA’s newly launched satellite altimeter system ICESat-2. We employed the method of matching ICESat-2 ground tracks with precisely surveyed terrain, represented by high-resolution DEMs. We expanded this classical approach and developed a least-squares approach to calculate the position of the altimetry profile that best agrees with the DEM. We found the highly-accurate DEMs of the McMurdo Dry Valleys, Antarctica, ideal for this research because of their stable landscape and rugged topography. We computed the 3D translation vector of 223 different laser altimeter profiles. Apart from the three unknown components of the translation vector the least-squares approach also delivers information about estimated errors.

A graphical representation of the translation vectors as a function of time reveals two distinctly different aspects of the uncertainty of the translation vector. For one it takes ICESat-2 about 10 seconds to traverse the Dry Valleys. Assuming that the calibration parameters of the ICESat-2/ATLAS observatory remain constant (neglecting small, high frequency changes) within a few seconds suggests to compute the mean and standard deviation of multiple DEM crossings. We estimated the uncertainty across-track to about 1.20 m and along-track to 0.90 m. One should bear in mind that these errors receive contributions from the DEM and ICESat-2.

The second aspect of the uncertainty of the translation vectors is to consider the entire time frame of the observations used in our experiments. This observation interval begins with the first official release of October 16, 2018 and ends by May 4, 2019. The across-track component spreads between 5 m and -9 m and between 6 m and -8 m for the along track component, respectively. We could not discern any long-term trend---it appears that the frequent calibration efforts keep ICESat-2 on track.