Figure 6 - Isometric view of basin model of Lake Garda obtained by integration of lake bathymetry with basin DTM 
and SCOP by INPHO . 
The bathymetric data collected along the track of the boat 
towing the sidescan sonar and published bathymetric data me- 
asured by the Italian Navy in 1966 (Istituto Idrografico della 
Marina, 1967). were used to generate a DTM for the lake bot- 
tom. To accomplish this it was necessary to effect a datum con- 
version as the bathymetry from the sidescan sonar survey was 
registered by GPS in the WGS 84 system while the 1966 data 
were referenced in the European datum (ED 50). To determine 
the conversion parameters six control points of known geodetic 
coordinates, thus referenced in the ED 50 system, were establi- 
shed around the lake and surveyed with a GPS instrument to 
record its position in the WGS 84 system. The Molodensky- 
Badekas model formulas were used to compute the transforma- 
tion parameters for datum conversion. The R.M.S. values were 
less than 12 cm. A comparison of the 1966 bathymetric data 
with that of 1994 indicated that in the northern part of the lake, 
near the lake's inlet, the bottom elevations were up to 20m 
higher while in agreement in the deeper parts of the lake. The 
discrepancies noted near the inlet are possibly due to sediment 
deposition over the 28 year period that separates the two sur- 
veys. 
For the final lake bathymetry the 1966 data were used for the 
shallow waters where no 1994 data were available. Perspec- 
tive scenes can be obtained by combining the elevation and 
the bathymetric data with the basin and the sidescan sonar 
images; an example is shown in figure 6. 
To integrate the sidescan sonar imagery with the onshore digi- 
tal orthophoto another datum conversion was needed from the 
Clarke 1866 ellipsoid to the ellipsoid of Hayford used in the 
ED 50 system. This became necessary because while the side- 
scan data are collected and georeferenced in the WGS 84 sy- 
stem they are projected in a cartographic plane by means of the 
1866 Clarke ellipsoid during their post processing. 
The same procedure was used with the image data: the side- 
scan sonar images of the lake floor were projected in the ED 
50 coordinate system and seamed with the digital orthophoto 
of the land to produce a continuous image of the entire lake 
basin as if the water were removed. The attachment of the si- 
descan sonar data with the shoreline as derived from the or- 
thophoto was not a simple matter because the programs that 
make the geometric correction of the acoustic data collected 
over a slanting range assume a flat floor. This may not be of 
great concern with regional data at sea or with gently sloping 
shores but introduces distortions when trying to fit sidescan 
sonar data to steep shores. Because of this the procedure of 
tying in sidescan sonar images to shorelines becomes almost as 
much of an art as a science. The technique relies on identifying 
tie points on the shore for which the sidescan data fit well and 
to wrap (and warp) the rest of the sidescan images around the 
neighboring shoreline assuming that the orthophoto is the geo- 
metrically corrected source image and the sidescan imagery is 
the uncorrected one. These procedures are executed by GCP 
works of PCI by means of a polynomial interpolation.The regi- 
stered sidescan image is then mosaicked on to the orthophoto 
image with a seamline coinciding with the shoreline illustra- 
tion of this for the upstream (northern) part of the lake is 
shown in figure 7. 
6.0 Filling of bottom imagery in shallow water 
In shallow water, with depths equal to or less than about 10m, 
near the shoreline where the vessel could not go there is no 
information on the lake floor. However the data gap can be 
filled by integrating available images from the first few chan- 
nels of the MIVIS hyperspectral scanner, with the sidescan 
data. Similar results could. be obtained with low altitude areal 
292 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
  
| 
Figur 
botton 
were I 
photog 
The D 
spectre 
the rac 
scanne 
the vi: 
infrare 
red bet 
redbet 
CASA 
it is lar 
to proc 
System 
southe: 
the dat 
Spectru 
the bot 
An exa