In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Yol. XXXVIII, Part 7B 
The relation between CDN and at-sensor radiances L 
[W/m 2 /sr/pm] is simply 
L = CDN/50 ( 2 ) 
Since at-sensor radiances include all atmospheric effects like the 
path radiance and the adjacency effect, the images will suffer 
from a blue hue, which increases towards the borders across the 
flight direction. Therefore the XPro workflow provides several 
options for atmospheric correction. The empirical “Dark Pixel 
Subtraction” and “Modified Chavez Method” will produce ap 
proximate ground radiances, which are no surface property 
because they will still depend on the actual solar illumination. 
In order to obtain an image product which is a surface property, 
in 2008 the “Atmospheric” option has been included in the 
workflow, which simulates the solar illumination and path radi 
ance, and then calculates reflectance calibrated digital numbers 
(.RCDN) (Beisl, 2008). The relation between RCDN and reflec 
tance R is 
R = RCDN /\0000 (3) 
An additional option is the “BRDF” correction which removes 
the cross track gradient caused by the reflectance anisotropy of 
ground surfaces. This correction produces homogeneous image 
strips calibrated to RCDN. 
2.3 Airborne data set 
In 2008 the DGPF started a project for the "Evaluation of Digi 
tal Camera Systems", which included assessments of the geo 
metric and radiometric accuracy performance as well as the 
performance in generating digital surface models (DSM) and in 
manual stereo plotting. The project was carried out in the con 
text of the EUROSDR initiative (Cramer, 2010b). A test site at 
Vaihingen/Enz near Stuttgart (48°56’N, 8°58’E) was chosen 
and medium format, large format, and hyperspectral cameras 
were flown by the manufacturers on several days (Cramer, 
2010a). The ADS40 S/N 30120 (SH52 type) was flown on Au 
gust 6, 2010 on two flight altitudes and two configurations 
(compressed and uncompressed image capture), as given by the 
DGPF directions (Table 1). 
Line 
GSD 
[cm] 
Start 
Time 
fUTCl 
Flight 
Heading 
n 
Sun 
Azimuth 
[°l 
Sun 
Zenith 
n 
IT 
RG 
[ms] 
IT 
BNIR 
[ms] 
Tilt 
config 
L6 
8 
0957 
90 
141.9 
37.3 
1.448 
1.448 
N+Bw 
comp 
L5 
8 
1002 
270 
143.7 
36.8 
1.452 
1.452 
N+Bw 
comp 
L4 
8 
1007 
90 
145.7 
36.3 
1.474 
1.474 
N+Bw 
comp 
L3 
8 
1012 
270 
147.7 
35.8 
1.476 
1.476 
N+Bw 
comp 
L1 
8 
1028 
270 
153.8 
34.6 
1.408 
1.408 
N+Bw 
comp 
L7 
8 
1035 
180 
156.6 
34.1 
1.800 
1.800 
N 
raw 
L2 
8 
1039 
270 
158.6 
33.8 
1.432 
1.432 
N+Bw 
comp 
L8 
8 
1046 
0 
161.3 
33.4 
1.800 
1.800 
N 
raw 
H5 
20 
1054 
180 
164.9 
33.0 
4.158 
2.908 
N 
raw 
H6 
20 
1100 
0 
167.2 
32.8 
4.482 
2.900 
N 
raw 
H4 
20 
1202 
90 
195.1 
33.0 
4.040 
2.790 
N+Bw 
comp 
H3 
20 
1213 
90 
200.0 
33.6 
3.866 
3.866 
N+Bw 
comp 
H2 
20 
1228 
90 
206.1 
34.5 
3.590 
2.340 
N+Bw 
comp 
H1 
20 
1240 
90 
210.8 
35.5 
3.908 
2.658 
N+Bw 
comp 
Table 1. Flight lines (L#, H#), solar angles, integration times 
(IT) for the colour bands, viewing directions (N=nadir, 
Bw=backward) and compression setting (comp=compressed, 
raw=uncompressed) for the Vaihingen/Enz data set. 
The flight project contained two different data sets: The first 
data set consisted of E-W flight lines at a lower level (800 m 
above ground, L#) and a higher level (2000 m, H#), corre 
sponding to a ground sampling distance (GSD) of 8 and 20 cm, 
respectively. The data was registered in compressed mode and 
was used for geometric calibration. The second data set with 
N-S flight lines having 8 and 20 cm GSDs was registered in 
uncompressed mode, which increased the radiometric resolu 
tion, to be most suitable for radiometric validation. 
The images, which viewed the radiometric test site (L3, L4, L7, 
L8, H2, H3, H5, H6) were processed with XPro 4.3 using stan 
dard settings to produce three standard products, namely at- 
sensor radiance (ASR), atmospherically corrected reflectances 
(ATM), and atmospherically and BRDF corrected reflectances 
(ATMBRDF). 
2.4 Ground measurements 
The radiometric calibration accuracy of airborne sensors is dif 
ficult to validate with absolute ground radiance measurements 
since cheap fiber spectrometers tend to drift in the NIR and so a 
stable absolute calibration for those spectrometers is difficult to 
obtain. Furthermore, due to changes in atmospheric transmit 
tance the solar irradiance changes rapidly. This means that the 
measurements have to be taken at the very time of overflight, so 
measurements at different targets cannot be taken with a single 
spectrometer. Therefore, reflectance measurements relative to a 
white calibration standard (e.g. Spectralon®) are performed, 
instead. This requires modelling the solar irradiance either for 
calculating the ground radiances from the measured reflectance 
data or calculating a reflectance product from the image data. 
Since the Leica XPro software provides a reflectance product 
when using the correction with “atmospheric” setting, the vali 
dation was done by comparing the XPro image reflectances 
with measured ground reflectances. This means testing two 
calibrations of the ADS sensor simultaneously (absolute radi 
ance calibration and reflectance calibration) against the Spec 
tralon calibration and spectrometer stability, i.e. possible devia 
tions cannot be attributed to a single source. However, it makes 
sense to test the reflectance product that is finally used. 
The radiometric test site contained tarps in four colours (white, 
blue, green, red, provided by RAG Deutsche Steinkohle), 3 
Siemens stars for resolution measurements, and a grey wedge 
tarp, all provided by the University of Stuttgart. The University 
of Stuttgart made ground measurements with a nadir looking 
field spectrometer, a camera-based goniometer, and two Sun 
photometers (Schonermark, 2010). 
The day was not optimal for radiometric tests, since high cirrus 
clouds (altitude > 3 km) passed by frequently. Thanks to an 
experienced flight crew, the radiometric test site could be im 
aged without direct cloud shadows except in line H6. However, 
the aerosol optical thickness (AOT) at 534.3 nm calculated from 
the Sun photometer data (Figure 1) shows strong haze effects 
during the time of ground measurements and flights. The AOT 
was varying between 0.4 and 1 for the different flights resulting 
in a change of direct solar irradiance by a factor of 1.8. Al 
though this is partially compensated by an increase of the dif 
fuse irradiance, without atmospheric correction, the overall 
image brightness will vary in this order of magnitude. Since 
ground reflectance measurements consist of two consecutive 
measurements of a target and a reference plate, even a small