Derivation of a global land elevation data set from satellite radar altimeter data for topographic mapping

Radar altimeter data from the 336 day cycle of the geodetic phase of ERS-1 has been processed over all land surfaces to produce a consistent, homogenous data set for topographic mapping purposes. Several processing steps were required to achieve this, the most important of which were waveform retracking and data filtering. No slope correction was applied to the data as this was found not to improve the accuracy. The original radar altimeter data set was 145 Gb in size and a more compact, higher-level product was required for land topography applications. Three higher-level products were created: 1) an unaveraged data set comprising elevation, position, backscatter and a small number of auxiliary parameters; 2) an averaged data set of quasi-regular mean elevations, position, and standard deviation at 5 are min resolution; 3) a 5 are min digital elevation model interpolated and extrapolated from the second data set. The accuracy of the radar altimeter data was investigated using six point data sets of both global and regional coverage, including airport runway elevations, national geodetic data, gravity station heights, and navigation beacon heights. A small number of erroneous elevations in both the radar and the point data sets were identified. After removing outliers with a difference of greater than 300 m the mean difference between the radar and point data was -0.5 m +/- 44 m full width at half maximum.