NASA Technical Reports Server (NTRS) 20000056934: Intercomparison of Laboratory Radiance Calibration Standards

5 Pages·0.36 MB·English

by  NASA Technical Reports Server (NTRS)

Intercom arison of laboratory radiance calibration standards
P .....
Betina Pavri, Tom Chrien, Robert Green, and Orlesa Williams
Jet Propulsion Laboratory, California Institute ofTechnology, Pasadena, California
Abstract: Several standards for radiometn'c bundle input was used to collect the comparison
calibration were measured repeatedly with a data. Sources were measured over a period of
spectroradiometer in order to understand how they several hours, and the results were compared.
compared in accuracy and stability. The tested
radiance standards included a NIST IO00W bulb
and halon panel, two calibrated and stabilized ANALYSIS
integrating spheres, and a cavity blackbody.
Results indicate good agreement between the Radiance Calculation
blackbody and IO00W bulb/spectraIon panel. If The spectroradiometer has an adjustable dynamic
thele two radiance sources are assumed correct, range based on an adjustment to the integration time
then the integrating spheres did not conform to their in the VIS/NIR, and changes to gain and offset
manufacturer-reported radiances in several regions beyond 1000 nm. However, since all data sets
of the spectrum. More detailed measurements are were collected without changing these settings, the
underway to investigate the discrepancy. DN levels are directly comparable. They can be
used to derive radiance curves for each source, if
INTRODUCTION one radiance source is considered to be the
The radiometric calibration of AVIRIS (Airborne standard. The new integrating spheres were chosen
Visible/InfraRed Imaging Spectrometer) is a process as the standards for the first part of the experiment:
which is constantly being refined. For the past measuring the lamp/panel radiance:
several years, an Optronics 1000W NIST-traceable
bulb and 12" Labsphere Spectralon panel (Figure l) .dnpanel/lamp(_,)
have been used to provide a radiometric standard for Lpanel/lamp=Lsphere dnsphere(_.)
laboratory and runway calibrations of the instrument
(Chrien et al., 1995). As specified by NIST, the
Fifty such spectra were averaged to produce the
lamp is positioned 50cm from the panel.
reported results for the lamp and panel.
aperture This radiance could then be compared to the
Jk radiance expected from the lamp/panel combination
I0°from
\ based on the manufacturer's calibration data for
vertical
lamp irradiance and panel reflectance (Figure 2).
lm
1000W bulb llamp•RpaneI
Lpanetthe°retlc=al n
,,,/_baffle
50cm 91
i,!,I
_pectralon panel
Figure I: AVIRIS calibration fixture
_4
There are many possible radiometric calibration
sources which might provide increased accuracy or
stability. Among these are stabilized integrating
I / calc,,Jlat,dr,_,nce um_g IlptlenlD 20201 4_7
spheres and calibrated blackbody sources. Two .4 ;...... calculated ra,'qlmce ulllr_ sphere 2_1' i
0 -----_ .... -_-- _-------_- --+_-_---F------+_ _ -_
new Optronics integrating sphere sources were 300 500 700 900 1100 1300 1500 1700 1900 2100 2300 2500
purchased and compared to the current standard in wlltvelenolh (rim)
the laboratory. An Analytical Spectral Devices full- Figure 2:panel radiance from sphere standard
range (ASD FR) spectroradiometer with a bare fiber
RESULTS
This is also true for the oscillations between 1_
The fit is poor. There are two explanations for this.
and 2,170 nm in the 700K data. However,_'
The first is that the panel and lamp were poorly signm levels are low, and the 70OK is the _':
calibrated by the manufacturers, andthe second is
temperature for which data is available foi_'_
that the reported sphere radiances are incorrect. portion of the spectrum (the higher radiance _'_v_:;
Since both spheres give essentially the same_resuit saturated the spectroradiometer). _.._._
for panel ra&ance, this would imply that the spheres The fit between the measured and _'_
share acommon calibration error.
data,forthe lamp/panel calibration case is ;__._
A blackbody source was selected to resolve the
that derived assuming that the spher_ _:i
ambiguity, as it should be free of the kind of water
accurately calibrated. Therefore, the disc_l___ _:-
absorptions that occasionally affect spheres and
between the sphere and lamp/panel calibrafiom _ _
panels used as radiance standards. All three sources
most likely the result of errors in the calibration of:
were measured using the same spectroradiometer.
the _heres.
Blackbody radiances were calculated using both the the agreement between the theoretical and
lamp/panel and a sphere as
_..ometric standards, measured blackbody radiances (using the }
The calculated blackbody
temperatures could in Figure 4: " . ;:?:
blackbody'equationa,ssuming an en_'vky of I
(Figur3e).
250 -.
- ---,m( _,_. t.,mm, |_o_
_oom I_uoot. mo,_m _t-mL
"°I-"2..,me,
L_O.) = 2_'hc2" 1
_.s ec_o.k_r1 "--'-"" "S "J
where:
c = speed of light
k = Boltzmann's constant
h= Planck's constant
I=
= wavelength of interest
¢= emissivity
1ooo_oo _=oo_oo _oo _soo _eoo_7oo_soo
•nM,.,_ fro) i
,oo ,_ , =-_ :_._-_=--_-_ Figure 4: sensitivity of theoretical blackbody _ to
_ _ _ _ ._'_ temperanme.
_so....... _Ii_ii_ ,"" I
I
.... |IIili "
_oo. -- ---- -I_i,_mmCsmmiiml) ,," I
!ii.:.._. ____. so....--.
[ II some issues. A _ error in blac_
...... IIII l
......----" -" _' temperature or emissi_ty most likelyaccountsif_.
I #_"-.---- 11 the small positive slope seen in the _"
I ' atmospheric water vapor absorption artifact from the
;I /L "_ different..curve. Further, note what is probably
'°'- l lamp/panel measurements at -1350 nm. It
so. _/r _ q_ i! introduces error into the radiance derivation,
:-..- causing the calculaterdadianceof the blackbody to
0- -,,-'__-" ,'7--'-=_ be underestimated
here.This does not explain
-,--_ ,. discrepancies in the integrating sphere data,
Figur3e:blackbodryadiancferomlamlOamlandsphere$. however, as this spectral feature is offset from the
discrepancy observed at 1300 nm in the sphere data,
The blackbody source saturated the long- andthere isnocorrespondingfeatureat 1700 nm.
wavelength end of the spectroradiometer at high
temperature,but good data were obtained I_m Radiance of Integrating Spheres
1000-1800rimforthe700,800,and I000K cases, Next, the radiance of the spheres was
andovertheentirespectrumforthe700K case. determined, assuming that the lamp/panelis
Note that1300nm and 1700 nm discrepancies accuratelcyalibrated. These spectrawere compared
appearonlyinsphere-derivesdpectrainFigure3. to the reported sphere radiances, in Figure 5.
Next,!he.discrepanciiensthe spherecalibrat/0n
,,0 ....... ,-_-----.p_,,ano_._ were cnecKea to estabhsh whether they matched'_6 _-
,e0 _ _ _, _ watervapororliquidwater(Figure7). :._'.
!.00 .....
,0./,.,% f ,,o_ .. ,,
tOO 0 i
,40 /: ,'" .....--_-- Y![_,:: : \ , ._ _:;,.
,o r | o..oJ, Ik :'! : I r I, J
" 2/0°'00 _ I _"o.,or' _l :,i,' "!_ '" IItI/ "'t/'
-- -- o.4ot 'h I ,; !
50o,oo.oo,,oo,.o,.o,.oo,,oo,,oo.oo.o|oo,o _;:: V '
i
Figure 5: rodianceof integralioaspheresbaaed on NIST
lO00W bglbamlapectmloapan_lamadard ;=,-:--: , : : _.___:_______.___._
.. ,. ,.. ,,.,..,...,.,._,.. ,_:_
are snmotber thaa themeasmed resalts,md that the -:" _**_ {.m) . ._..,,
discrepancies for both spheres track one another
Figure 7:wanxmb_ ofwater (vapor and liquid) :::..
(note discrepancies at 1300 nm and 1700 ran).
Since the two spheres were calibrated by the This was done to establish whether the problem is_
manufacturer at the same time, this may be an adsorbed water in the sphere surface, or"
indication of a problem with the manufacturer's atmospheric water vapor in the fight path. Spheres
calibration routine, at least for _ day. High- may have significant water vapor signatures due to ::
frequency "noise" at I000 nm is most likely due to the large atmospheric path length (due to multiple'_"
low response of silicon detector (near its cutoff bounces in the sphere). A quick comparison _of
wavelength) !n the spectroradiometer. curves to the wavelengths at which '_ the
discrepancies occur indicatetshatwater vapor istbe
The disagreement between measured and reported mainfactor.Figure8 comparestheabsorptio_in
sphere radiances can be examined by displaying the thesphereradiancetoModtran-derivedwatervapor
results as apercent difference (Figure 6). absorptions.
' ................i.....,..o|.....
25% ,
::::::::::::::::::::::::.......,
!
I ::::::::::::::::::::::::::: :I
,.:o.:::::::::::::_::: .., :_._._:
.._/ I f: _,
.s% ...................... -_-¢-?,-I
-20% _ _ %cM_mnee- l
...... 2_1% ¢lllfwli'wl_ ................
-25% , , , J y , I , i I . /
"" I
300 SO0 ?00 O00 11001300 150017001_0G:1t0025002S00 |H S00 700 Nil, 11@4 1"00 ll_) 1100 1100 1_IN l$00 _I_I
wlvelenilll (mill) _um_olOalltk (am)
Figure6:percentadg_eereace_twe_acalcutaua_ndm_orua
Figure 8: sphere ra_Bance compared to a_u_al_Cri¢
radiance for integratin_gh_res tranamission, showing influence of water vapor. Arrows
indicate discrepancies not explained by water vapor
Figure6indicatedsiscrepancieosfup to20% from absorptio_
thereportedradiancevaluesatthelongwavelength
endofthespectrum.The osciIlat/ofnrom-15% to So, water vapor apparently accounts for most of
+20% from 2100 nm to 2300 nm isespecially the discrepancy, with the exception of two regions
troubling. of the spectrum. Since radiometric tests are done
under ambient conditions, water vapor absorption
-%
will be a critical (and variable) factor in radiometric
Further, spectral position .is repeatable ,_m
calibration if spheres areusedasthe standard.
measurement to measurement, though absolute_
are not, due to changes in positioning betw_
DISCUSSION
lamp and fiber during the period of meas_t'..
Spectral Calibration:
(both were handheld), andthe fact that the lan_'_i!;
It is possible that some discrepancies might result if not fill the fiber's FOV ":_$_
the spectroradiometer were experiencing a sl/ectral • %':_rL
shift during measurements. In order to establish Verification of Blackbody Temnerature: ___
whether the instrument was spectrally calibrated, a
The assumpaon of good calibration for the lam_51
mercury vapor lamp was observed as part of the
panel can be checked by independently verif_in'g'_g_ •
test. A series of 100 lamp observations was made
blackbody temperature setpoints, assunang the_i_
in order to accurately determine the spectral offset of emissivity of the blackbody was very close to 1.0. >
the instrument at each emission line. The measured
Microsoft Excers Solver function was used to
lines were compared to some of the strongest minimize the between
reported emission line.s for the lamp (Figures 9 and functions: the ratdioiscorfeptahnecyDN values fortwthoe 80ra0tKio
IO). The results m Figure 9 indicate that the and 1.000K ASD measurements of blackbody D'/i_,i__
instrument is speccaRy calibrated to within 2 nm in ana me ratio ol_the theoretical radiance valu_ -_.
the visibl_ region of _ _. + two te_ values (T_high _,and" .T_....i__;:_
'.-; " "_ : " " _. : T_high andT_low _:allowed to v_, _,in1__'_
to be was' sum-s ared*
30004
_,..,_,,,,] difference between the two ratios at "eia/li,_-,
- ..... _,_ _ j wavelength. "_ ':
25000
Using all points, with 1000K and 800K_
20000 specified as start points for T_high and T_low, the _
results were: .?_
1SO00
T_high 1000 K 998.35759 K
10000 I ori_inal temperature final temperature _
T_low 800 K 798.118056 K
sum-squared of differences I 116.945067
5000C_l
If the apparent water vapor absorption feattLre
540 542 $44 $46 S4S 550 between 1300 and 1500 nm ('Figure 4) is deleted
wavelength (am) from the analysis, the program generated this result:
Figure 9: spectral calibration test, visible region.
An IR spectral calibration is shown in Figure 10. T_high 1000K 995.813536 --
i orig_inaltemperature final temperature
Spectral calibration of the speetroradiometer appears T_low 800K 796.597793 ,::_
to be better here than in the viable.
sum-squared of differences [ 107.290529 _ •....
This results in somewhat better agreement ('as
400 measured by the sum-squared of differences), and
slightly lower temperatures than before.
3oo- I _-..... .m:tm,_m ;
The accuracy of this method was found to be highly
" i.... IdEIqCUPiYt _04 m
dependent on the starting values chosen for Thigh
250- _,_ ,.... MEN:UI_,_O6 I
and T_low. Starting close to the correct values is
c200 : _ERGURYI_
critical for achieving a sensible result; there
probably exist many local minima to this function.
'22 I The sum-squared of the differences between the two
ratios for the final values can be used as a guideline
for checking the progress of this process, but this
criterion alone is not sufficient tOcheck the results.
1880 I|8S 111110 1095 1700 170S 1710 The residuals between the measured and theoretical
wavelength (rim) radiances would need to be used as well -- so the
Figure !0: Spectral calibration test. IR region process is iterative.
CONCLUSIONS
Several radiance standards were compared against
e another using a portable spectroradiometer.
• results of the testing indicate that the current
AVIRIS standard, a 1000W N"[ST-traceable bulb
used in combination with a Spectralon panel, ,is the
most versatile and accurately calibrated o_" the
systems tested. The well-calibrated blackbody may
provide a check on the radiance of the bulb/panel,
but cannot cover the entire spectral r__{_,eobserved
by AVI]_S (400-2500 nm). water vapor
absorptions explain most of the discrepancies
between the calculated and manufacturer-reported
integrating sphere radiance. However. this does not
change the central result for our cafibration
purposes: these integrating spheres are not a good
primary standard for absolute radiometric calibration
ofpyperspectnsaystems.T .im
tesu_amay provemore usefulifthe_ _":-
observedaredueto amanufacturer c.ah'br_ion eaor, ......
and.if they are stable after cah'bradon is completed.
Further tests are underway to determine the stability
of all the systems discussed in this paper.
Acknowledgments: This research was carried out at the Jet
Propulsion Laboratory/C_.alifomIinastituteof Technology,
Pasadena, California, under conOm_t with he National
Aeronautics and Space Administration. We are grateful to
Bruce Kindel for helpful comments on this paper.
REFERENCES
Chrien, T.G, Green, R.O., Chovit, C., and Hajek, P. (1995),
New calibration techniques for the Airborne
VisibleHnfrared Imaging Spectromet/er (AVIRIS). In
Summaries of the Fifth Annual JPL Airborne Earth
Science Workshop, JPL Pubi. 95-1, VoL 1, Jet
Propulsion Laboratory, Pasadena, CA, pp. 33-34.