The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
219
The reflectance curves have been transformed into at-sensor
radiance by using MODO 10 , the graphical user interface to
MODTRAN 10 .
A median radiance level is retrieved my averaging the highest
curve (20% chlorophyll) with the lowest curve (i.e. 60%
chlorophyll).
In order to run the model, it’s necessary to derive the
application driven requirements in terms of center wavelength,
radiance, SNR, spectral resolution and noise-equivalent delta
radiance. Bands are defined by considering the most interesting
parts of the reflectance curves in Figure 6; the most demanding
portion of the curve is the one corresponding to the red edge
(i.e. between 680 nm and 750 nm). Therefore 4 bands are taken
here each one with a spectral resolution of 5 nm in order to
detect the slope of the red edge. Other 6 bands are considered
before and after the red edge with a spectral resolution of about
10 nm.
Table 3: Vegetation (Chlorophyll Content - Red Edge)
requirements.
À.R
|nni|
À'C
|nml
SSIr
|nm]
SSIc
|nm|
GSDr
|m|
GSD<
|m|
400.2
405.1
10
10.35
3.65
3.65
450.4
445.8
10
9.89
3.65
3.65
499.9
504.3
10
10.18
3.65
3.65
550.9
552.6
5
5.17
3.65
3.65
559.9
603.4
10
9.10
3.65
3.65
650.9
650.9
5
2.86
3.65
3.65
681.7
681.7
5
3.28
3.65
3.65
702.4
702.4
5
3.57
3.65
3.65
728.9
728.9
5
3.96
3.65
3.65
749.6
749.6
5
4.27
3.65
3.65
Lu lnm|
SNRr
SNRc
NeAL«
NeALf
F
400.2
404
745
le-4
6.7e-5
-
450.4
70
931
8e-4
6.0e-5
-
499.9
45
831
lle-3
5.9e-5
-
550.9
37
578
18e-3
1. le-4
-
559.9
31
803
16e-3
6.0e-5
-
650.9
30
403
12e-3
8.7e-5
-
681.7
31
399
10e-3
7.6e-5
-
702.4
29
414
16e-3
1.2e-4
-
728.9
84
483
7e-4
1.2e-4
-
749.6
551
552
2e-4
1.9e-4
-
The radiance requirement is defined by averaging the maximum
radiance curve with the minimum radiance curve. The minimal
difference of chlorophyll the scientist is interested in is 2%; this
corresponds to the noise equivalent delta radiance (NeAL), that
is, the difference in radiance between two radiance curves,
which differ for 2% chlorophyll content. Finally, we could
define the SNR requirement by calculating the ratio between the
medium radiance level and the NeAL. The requirements are
shown in Table 3: Vegetation (Chlorophyll Content - Red
Edge) requirements.
The results of the simulations are shown in Table 4, and
represented by the variables with the c subscript. It is apparent
how all the requirements have been meet with a tolerance of
about 5% in all cases. The SNR makes an exception; in fact all
the model SNR results are much higher than what requested. It
does imply that the NeAL is actually one order of magnitude
smaller than the requirement. In order words, within the limits
of the applicability of a linear assumption, that this instrument is
potentially able to distinguish chlorophyll content with an
accuracy of better than 0.2 %.
The binning pattern for such an application is shown in Table 4.
Table 4: Binning pattem for the chlorophyll/red-edge
application.
R
1
2
3
4
5
6
7
8
9
10
hirst
42
104
163
198
223
243
253
259
266
271
Last
59
115
170
200
226
243
253
259
266
271
4. CONCLUSIONS
Specific scientific requirements might differ from the ones a
hyperspectral pushbroom spectrometer has been designed with.
An optimization algorithm is here presented. Such a model is
based on both the SNR equation and the basic instrument
electrical and optical parameters. The main goal of model is to
provide a sensor configuration in terms of integration time and
binning patterns in order to let the sensor meet the specific
application requirements. Additional solutions are also
discussed whether the instrument variables cannot be optimally
tuned. Two case studies are therefore presented. In the first one
a generic scenario has been used to define the default instrument
requirements in terms of spectral and radiometric parameters
and the corresponding nominal sensor setup is defined; a few
requirements can be met only if a special filter is used. The
second case study deals with a scientific application, that is, the
identification of at least 2% differences in chlorophyll content
within the optical signal generated by canopy. Results, errors,
and binning patterns have been presented.
This optimization tool can be easily adapted to any sensor and
independently from any kind of platform (i.e. airborne and
spacebome). Its main advantage consists of using as good as
possible the programmability functionalities of current
hyperspectral systems.
References from Journals:
1. J. Nieke, K.I. Itten, W. Debruyn, and the APEX team, ’’The
Airborne Imaging Spectrometer APEX: from concept to
realization,” in Proceedings of 4 th EARSel Workshop on
Imaging Spectroscopy, Warsaw (2005)
2. J. Nieke, M. Solbring, and A. Neumann, “Noise
contributions for imaging spectrometers,” in Applied Optics,
Vol.38 No. 24 (1999)
3. I. Baarstad, T. Loke, and P. Kaspersen, “ ASI - A new
airborne hyperspectral imager,” in Proceedings of the 4 lh
EARSel Workshop on Imaging Spectroscopy - New Quality in
Environmental Studies, Warsaw, Poland (2005)
4. C. O. Davis, J. Bowles, R. A. Leathers, D. Korwan, et al.,
“Ocean PHILLS hyperspectral imager: design, characterization,
and calibration,” in Optic Express, Vol. 10, No. 4 (2002)
5. D. Schlapfer and M. Schaepman, “Modelling the noise
equivalent radiance requirements of imaging spectrometers
based on scientific applications,” in Applied Optics,
OSA41(27):5691-5701.
6. R. O. Green, “Spectral calibration requirement for Earth
looking imaging spectrometers in the solar-reflected spectrum,”
in Applied Optics, Vol. 37 No.4, 683-690 (1998)
7. P. Mouroulis, D. A. Thomas, T. G. Chrien, V. Duval, R. O.
Green, J. J. Simmonds, A.H. Vaughan, “Trade Studies in
Multi/Hyperspectral Imaging Systems Final Report,” in Jet
Propulsion Laboratory files (29 October 1998)
8. . R. B. Gomez, “Hyperspectral imaging: a useful
technology for transportation analysis,” in Optical Engineering
41,2137-2143 (2002)