Downloaded From: http://proeeedings.spiedigitallibrary.org/ on 10/22/2015 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx 
4. EVALUATION OF THE DATA SETS 
In this chapter we examine the 3D points classified as seabed at different depths. We also analyze the vertical accuracy 
of the 3D points with respect to the depth obtained by echo sounding in Section 4.4. In order to eliminate the effects of 
turbidity we give the result in terms of fractions of Secchi depth rather than in actual depth, by dividing the actual depth 
by the Secchi depth. 
In order to characterize the seabed depth we use echo sounding data, which were measured by the Federal Maritime and 
Hydrographic Agency of Germany over the last 20 years. The horizontal distance between neighboring points of these 
measurements lies between approximately 5 and 50 m. A number of recent ship-based measurements, carried out in a 
few small areas, verified that the morphology from echo sounding is in general still up-to-date. Thus, the older echo 
sounding data which cover the entire study area (except the very shallow water zones) can be considered as reference 
data for our analysis. 
To evaluate ALB data sets according to Order la/lb of IHO Standards for Hydrographic Surveys (S-44) [19], a grid with 
a cell size of 5 m was generated from the data using a linear Delaunay interpolation method as implemented in Matlab 
[20]. The Secchi depth data were also interpolated, with a grid resolution of 10 m. Dividing the depths of the interpolated 
echo sounding data by the corresponding interpolated Secchi values leads to a normalized seabed depth grid, in which 
the depths are represented in relation to Secchi depths. Since the very shallow water zones were not accessible for 
vessels, the normalized seabed depth between 0 and approximately 0.2-0.5 times the Secchi depth are incomplete. 
4.1 Point distribution 
The point distributions of the 3D points classified as seabed of each campaign are shown in Figure 2. 
The seabed could be observed approximately up to 5 m water depth by the Riegl VQ-820-G sensor in the first campaign 
in November 2012. For the second and third campaign the same sensor Chiroptera with the same flight altitude was used. 
However, due to the different turbidity conditions during data acquisition the detected depths are different. In the second 
campaign in September 2013 the laser reached up to approximately 10 m, but in the last campaign in May 2014 the 
penetration depth was about 7 m owing to the poorer average Secchi depth in comparison to the second campaign. 
Proc. of SPIE Voi. 9638 96380Z-4