The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part Bl. Beijing 2008
268
than ALS parameters (Hyyppa et al. 2005). Thus, the quality
analysed for flat surfaces may be a neutral way to perform the
analysis. Concerning nationwide laser scanning, rapid, cost-
effective field surveying methods are needed to perform such an
analysis.
As a result of the test project final flying and scanning
parameters were further defined for operative nationwide laser
scanning. Accuracy requirement for the new 2 by 2 m 2 grid
model was set to be better than 30 cm (RMSE). 20 000 km 2 will
be operationally surveyed using airborne laser scanning in 2008
by the NLS and private companies.
The GPS reference station has been set up on the nearest
benchmark or a virtual reference station has been used.
According to Hàkli (2004) the RMSE of VRS RTK is 2 cm in
xy and 4 cm in z. Bilker and Kaartinen (2001) give RMSE
accuracies 1 cm + 1-2 ppm in xy and 2 cm + 2 ppm in z in their
RTK GPS report.
Xy-locations for planimetrie accuracy estimation: Reference
points were measured with Leica SR530 Real Time Kinematic
(RTK) GPS. Horizontal accuracy of the RTK measurements
was verified to be about 0.015 m and vertical accuracy 0.02 m
in another study (Bilker and Kaartinen, 2001).
2. MEASUREMENTS
2.1 Airborne laser scanning
Two acquisitions were planned for the test: first performed by
Optech ALTM 3100 in late 2006 (under leafless, snow-covered
(less than 5 cm conditions), and the second performed by
ALS50-II in early spring 2007 (after the snow melt, leaf-less
season).
An airborne laser scanner Optech ALTM 3100 was used for
scanning the northern part of the NLS test area and performed
by TopScan GmbH on behalf of the SITO group, a Finnish
Consulting Engineers Oy. Flights were carried out on the 21
December 2006. Flying heights were about 2000 m above
ground. Aircraft was a Cessna 404. The project size was 700
km 2 . Parameters were as follows: Pulse repetition frequency
(PRF) 50 kHz, maximum scan angle 20 degrees, scan frequency
24 Hz, planned swath width about 1380 m. All laser points
including strip adjustment were supplied to the customer and
the accuracy analysis was carried out using these calibrated
points.
An airborne laser scanner Leica ALS50-II was used for
scanning the southern part of the NLS test area around the city
of Salo in south-western Finland. The NLS rented the scanner
from the FM International Oy and operated it by themselves.
Parameters were as follows: Pulse repetition frequency 52 kHz,
scan angle was ±20 degrees, scan frequency 45 Hz, planned
swath width 1600 m. The flying height was 2200 m above
ground. The resulting point density was 0.5 points/m 2 .
Eastern part of the area was also flown from 4750 m and also
partly 500 m flying heights were used. The NLS operated the
twin engine Cessna 40IB on April 26 - 30, 2007.
Strip adjustment was done with the TerraMatch software at the
FGI. This software solves for systematic errors in position,
orientation and mirror scale using flight trajectories, classified
surface laser points linked to the trajectories with time stamps
and additional known points if they are available. First we used
the heading, roll and pitch shifts and mirror scale values for the
whole project area. These values were provided by the NLS.
After that the height shift and roll values were solved for
individual flight lines and strips were corrected.
2.2 Reference points
Terrain elevations for various land cover types: In addition to
the RTK GPS measurements in open environment, tachymeter
measurements were made in the forests. The RTK GPS have
been used for the starting point measurements for tachymeter in
the areas where benchmarks were not available. These ground
reference point measurements were made in 2003 and 2007.
Terrain elevations from hard targets to cover the collected area:
Mobile VRS RTK measurements were made on May 24, 2007
to collect reference points for airborne laser scanning in Salo
area in Finland. A GPS antenna was assembled on the roof of a
van. Co-ordinates were registered every 2 seconds which
corresponded to 20 to 30 m point distance depending on the
speed of the car. More than 1400 points were measured during
the day. Road numbers 52 Pernio - Salo, 186 Salo - Kisko and
1870 Kisko - Kitula were measured. Trees and cuttings
obstructed the GPS signal and caused interruptions in
measurements. Open field areas were excellent for this kind of
mobile measurements.
6700000
^ 6695000
% 6690000
Z 6685000
cconnnn
.>\ •
\ ' s *
) v
/
280000 300000 320000
East
Figure 1. The geographic location of the mobile VRS-RTK
points used for laser scanning height accuracy
analysis, N=251 points.
2.3 Analysis
The comparison between the observations and reference points
was made as follows. Laser points were selected so that the
reference point and the compared laser point were within 0.5 m
distance from each other. To increase the amount of compared
points also distance less than 1 m was used for comparison.
If there were one or more laser points within the predefined
distance (e.g. 50 cm) from the reference point the nearest laser
point was selected and the height difference between the laser
point and the reference point was calculated. In the earlier study
of Ahokas et al. (2003), it was found that the interpolated height
value, mean height value and the nearest laser point height
value gave about the same height results in the comparison
analysis. Laser points in various land cover classes were
analyzed in this way.
The road point analysis showed that the registration of laser
points from black asphalt was sometimes difficult. There were
reference points available on the road in the open field
environment but the corresponding laser points were missing
due to the weak return signal.