Estimate of the ASTER GDEM regional systematic errors and their removal

Abstract

The aim of this article is to analyse and estimate the systematic errors of the Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model (ASTER GDEM), assuming the existence of systematic influences due to the erroneously determined altitude datum. The assumption is based on the identical sign for the majority of differences between the control point elevations and the elevations derived from GDEM. The question is how to model a systematic influence and whether such a model can improve the global GDEM accuracy. The error analysis comprises two unbiased test areas (flat and hilly areas), where one uses the elevations of the control points as conditionally correct values. Three methods of estimating and eliminating the systematic errors are presented: a constant-systematic-error model, a variable-systematic-error model by first-order polynomial and a variable-systematic error model by second-order polynomial. The accuracy analysis concerning the original and calibrated GDEMs indicates a significant accuracy improvement in the case of the calibrated methods for both test areas, especially for the flat area.