The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
316
consequently lower depth for that point relative to the adjacent
topography. He has reported an error order of 1 meter at an
average depth of 22. Radar sensors have the advantages of data
collection over night and also turbid water depth computation.
Hyperspectral airborne images like: Compact airborne
spectrographic imager (CASI), Airborne visible-infrared
imaging spectrometer (AVIRS), Advanced airborne
hyperspectral imaging systems (AAHIS) and Digital airborne
scanner (DIAS) are mostly used in biological investigations
(coral mapping, chlorophyll estimation, identification of other
marine vegetation, water temperature) and there are some
activities on bathymetry information extraction from CASI data.
CASI sensor has the ability to acquire 288 bands and gives the
ability to the user to select the bands which are suitable for the
bathymetric information extraction for a particular area of
interest. Choosing the right wavelength makes it possible to
calculate regression between depth and reflectance for clear and
turbid water. But due to the confusing effect of variable depth
on bottom reflectance, the computed depth measurements have
limited accuracy of order 1 meter for up to the depth of 22
meter, which does not satisfy bathymetry depth measurement
standards. For deeper water, lidar sensors not only have the
ability to measure the deepness down to two or three times the
Secchi depth at 532 nm (equals to approximately 50 meters in
clear water) but also have higher spatial resolution, below
surface object detection larger than lxl m2, high data
acquisition rate per m2, acquires direct 3D position and need
less data post processing and rapidly is available on the
emergency situations. The planimetric and vertical accuracy of
Lidar sensor is dependent on the flying height. Currently Lidar
data accuracy satisfies the bathymetric standards. Detection of
the zero depth which is river banks and its displacement are
very time consuming and expensive task by traditional
hydrographic methods and can be detected by Lidar sensors
more rapidly with lower costs. For the second factor which was
coast or river surface and banks topography, only
photogrammetry stereo images and Lidar sensor are capable of
the topography extraction and the remaining sensors produce
2D data. A comparison between Lidar and photogrammetry is
demonstrated by Lane (Lane et al., 2003). He has given more
priority to the photogrammetric approach in the flood extent
extraction comparing to the Lidar sensor. But if one could focus
on his results, it would be apparent that: 1) For flood extents,
there is negligible difference between accuracy of data derived
by Lidar and photogrammetry, 2) Photogrammetric method has
problems for the water surface topography but against to this
Lidar is capable of supporting those kind of information and
also for large flood extends, photogrammetric approach needs a
lot time for the data processing but Lidar data can be delivered
to the user very rapidly on emergency needs. Bates (Bates et al.,
2002) and Cobby (Cobby et al., 2001) have done investigations
on vegetation height extraction near from Lidar data to improve
flood modeling which can not be detected by other remote
sensing sensors. Pereira (Pereira et al., 1999) recommends
using Lidar sensor instead of photogrammetric image due to its
rapid and cheaper product. Man-mage objects like, bridge and
building are particularly favored because provide a tool for
Lidar calibration providing a check for horizontal and vertical
alignment.
Ackermann (1999) gives an overview to present status and
future expectations of airborne laser scanners. Baltsavias
(1999a, b) discuss the basic formulas and existing systems and
Wehr and Lohr (1999) presents an introduction and overview to
airborne laser scanners. Mohammadzadeh et al. (2006) gives a
brief review to the some of the exiting research works in
different applications of lidar technology. Hydrographic Lidar
calculates the water body depth in shallow rivers and coastal
areas using the time difference of blue-green channel and
infrared channel reflected from the sea bottom and the water
surface respectively. Schmugge et al. (2002) has made a brief
survey on the past remote sensing solutions used in
hydrological problems. Cunningham et al. (1998) and Irish et al.
(2000) give a good overview on the airborne Lidar hydrography
program. Among all the operating hydrographic Lidar sensors
SHOALS (Scanning Hydrographic Operational Airborne Lidar
Survey) is an airborne Lidar system in the world that collects
both hydrographical topographical measurement in a single
survey (Guenther et al., 2000). Before flight, some calibration
processes are carried out by the instrument designing company
and therefore some researchers have focused their research on
fundamental aspects such as: laser scanner calibration (Adams,
2000) and (Wagner et al., 2006), accuracy improvement
(Latypov, 2002), strip adjustment (Bretar et al., 2004), noise
reduction of lidar signal (Fang and Huang, 2004), lidar
backscatter modeling (Fochesatto et al., 2004), lidar beam
alignment (Latypov, 2005), and stability of laser swath width
(Luzum et al., 2005). Also there are other calibration activities
needed to be performed before the flight starts in the field. The
position shift among laser scanner, GPS and IMU should be
measured accurately to apply the spatial shift among them. Also
scan rate, GPS/IMU data acquisition rates are not the same and
should be synchronized. The maximum detectable depth by
laser scanner is varying according to the water turbidity and
small particles in the water. The white calibration disk should
be used to calibrate the laser backscatter from different depth of
the water body. Scan rate, flight height, flight lines, designing
control points to mount GPS instruments and project cost
should be determined before performing flight over the region
of interest. During the flight, the human expert should check the
overall accuracy of the data to avoid large and unexpected
systematic errors. The coverage between acquired data should
be monitored to avoid data gaps. Afterwards all the acquired
data should be processed simultaneously to convert raw data set
to LAS or ASCII format readable by lidar processing software.
The primary effort in all the hydrographic applications is
transformation of the Lidar point cloud into a desired projection
system. Outliers can be filtered out to have more realistic
dataset. Then advanced image processing algorithms are
applied according to the user needs and application nature. The
intensity information is interpolated or in some cases is
estimated using other source of data to produce raster image.
The use of intensity derived image and optical images makes it
possible to better recognition of the outlines of the coastline and
nearby standing objects. Various approaches are developed in
each specific hydrographic and oceanographic case to obtain
required value added information such as: rapid high-density
measurements of the coastal zone (Saye et al., 2005); track
movements of sand placed for beach nourishment (Shrestha et
al., 2005); reveal linkages between changes in offshore
bathymetry and shape of the shoreline (Thoma et al., 2005);
flood prediction (Webster et al., 2004); water surface (Hwang et
al., 2000) and bottom reconstruction in bathymetric lidar (Pillai
and Antoniou, 1997) ; coast or river surface and banks
topography (Bates et al., 2003), (Charlton et al., 2003);
Vegetation height (Bates et al., 2002), (Mason et al., 2003),
(Cobby et al., 2003), and (Cobby et al., 2001); effect of man
made objects on tidal rivers (Gilvear et al., 2004); tidal
channels geomorphology (Lohani and Mason, 2001). In the
following part a synthesized discussion around important
exiting problems in lidar data processing and possible solutions
will be discussed.