The following example shows how an editing function could be performed on the DEM extracted from a SPOT satellite stereo pair using the EASI/PACE SDEM program in the Satellite Ortho and DEM Package.
The user opens an ImageWorks session with one 8-bit channel and one 32-bit real channel, and loads the satellite image into channel 1 and the DEM into channel 2 (the 32-bit real channel). The PCT editing panel is opened to select the "Stepped" pseudocolour table since this better shows elevation data.
The DEM editing panel is opened and the user selects channel 2 as the DEM to edit. Note: the DEM editing panel fills in the Failed and Background values automatically from the channel descriptor that SDEM made. In this case the failed value is -10 and the background -20.
The user notes that there are a number of urban areas which were extracted poorly; not only are there failed values, but there are a lot of bad/noisy values. The user masks these areas and sets the entire area to the failed value using the "Fill using Value" function, then interpolates these areas using the Interpolate "Under Mask" option. Then the user clears the mask.
The user also notes a large area which failed to interpolate due to cloud coverage in a mountainous area. Since the area is so large and the relief so complex, interpolating this area would be impossible. So the user decides it should just be ignored. The user masks out this area and sets it to the Background value using the "Fill using Value" function.
The user notes that the numerous lakes in the area are not flat, and in some cases are full of bad values and failures (typically since lakes themselves have no features for the automatic DEM extraction to match on). For now the user ignores the lake areas since this will be fixed at a later stage.
At this point the user selects the "Remove Noise" function for the entire DEM and uses it twice, followed by the "Interpolate" for the the "Entire DEM" (which interpolates all the failed values so none are left) and the "Smoothing" function twice. This creates a good DEM except for lake areas.
Examining the corresponding SPOT image (which was originally loaded into channel 1) the user notes that lakes in a SPOT image have very low grey levels, typically 10 or less. Using the "Modelling..." tool (only available in ImageWorks) the user enters the following equation:
if (%1 <= 10) %2 = -10
In other words, if a pixel in channel 1 has a grey level of 10 or less, make
the corresponding pixel in channel 2 (the DEM) -10, or the failed value.All the lakes have now been set to failed values in the DEM, removing the original DEM values and interpolated values. The user now locates a large lake and creates a mask for the lake using the "Fill Failed@Cursor" button. Looking up the elevation for the lake, the user enters this value for the "Fill Using Value" function, which sets the entire lake to a nice flat surface at the correct elevation. The mask is then cleared using the "Clear Mask" function and the next large lake selected.
Finally the user notes the dozens (hundreds) of little lakes left over, still identified by failed values. Lacking the time to locate and fill each little lake, the user opts for the following method: the "Fill all Failed" button is used to create a mask covering all the remaining lakes; the Interpolate "Using Mask" option is then used interpolate values for the failed values in the lakes by using the shore of the lake; then the "Fill Each Polygon with Polygon Average" button is used. This last step identifies each lake as a polygon and uses the average elevation for the lake as the elevation for each pixel in the lake (this makes it flat and approximately at the correct elevation).