9st
ra
he
Erik de Min
3.2 Correction for strip deformations (bending deformed boards into flat ones)
The above described strip adjustment does not always yield satisfying results. An example is given in figure 10. The
residual profiles illustrate that a systematic effect with a wavelength of about 15 km obviously remains in the residuals
after strip adjustment. This can be explained by a deformation of the strips. Thus, they cannot be treated like stiff
boards. The various deformations that appeared in the laser altimetry projects of the Survey Department can be
described by a cross strip parabolic deformation, a periodic effect in along-track direction and even torsions of the
whole strips. These deformations can differ for every individual strip. The error sources are mainly unknown.
Figure 10: Residuals showing a systematic effect after strip adjustment
In (Boon, 1999) it is demonstrated that strip deformations are responsible for disturbances in the results of strip
adjustment. Figure 11a illustrates, for example, the undesired effects of cross strips parabolic deformations on the
resulting strip adjusted block. In figure 11b the roll and offset parameters that are computed from the corresponding
strip adjustment are depicted. They are up to several metres. The offset values cross the zero axis at strips where
reference measurements are available. The results improve dramatically when cross strips are added. Hence, whenever
adding cross strips yields large changes in the estimated strip adjustment parameters a, ^ and c, strip deformations are
probable. This way, cross strips are used as indicators for possible occuring deformations. They are not primarily used
to correct the data.
From the height differences of tie ‘points’ in the overlapping areas, it is not possible to say whether the additional
systematic errors originate from the left or right strip, or even from both strips. Therefore, a different approach is
chosen, which does not make use of strip overlaps. Instead, the heights of the complete strip are compared with
reference heights of "TOPhoogteMD'. This is the former height dataset, covering the Netherlands with an average
density of 1 point per hectare. Although the point noise of this dataset is about 50 cm, it can be assumed that systematic
errors are absent over several kilometers. It is therefore supposed that systematic effects in the differences between
"TOPhoogteMD' data and altimetry data are caused by the latter.
Before strips adjustment
mue re ner
After strips adjustment
Figure 11a: Influence of across-track parabolic Figure 11b: Estimated roll and offset parameters
deformation on the results of strip adjustment in the strip adjustment for a block of 23 strips
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 235