Augmentation of vertical accuracy of digital elevation models using Gaussian linear convolution filter

We employ local levelling data to assess the global digital elevation model (DEM) from Shuttle Radar Topography Mission (SRTM3) over Khartoum State area. A linear convolution low-pass Gaussian filter is employed to reduce noise inherited in the DEMs. The main sources of noise are likely propagated during data collection due to the different impacts that may cause sensor noise. The systematic errors in the differences between the DEM-based and levelling heights are removed by using 4-parameter and third order polynomial models. Both models have improved the vertical accuracy of SRTM3 up to 3 metres after filtering and for only a decimetre without filtering.