Version 2.9 (December 2017)
Written by: G. FILACCHIONE (INAF-IAPS, Rome), E. AMMANNITO (INAF-IAPS, Rome)
Document Change History
===========================================================================
Change Date Affected Portions Checked by Approved by
===========================================================================
Version 1.0 January 2011 A. CORADINI
(INAF-IAPS,
Rome)
Version 2.0 October 2013 Chapter 6 : added details M. GIARDINO M.C. DE SANCTIS
about the responsivity (INAF-IAPS, (INAF-IAPS,
formula. Rome) Rome)
Chapter 7 : added some M.T. CAPRIA
explanation about usage (INAF-IAPS,
of internal calibration Rome)
Chapter 8 : added details
about the dark current
subtraction procedure,
corrected CALIB directory
file names.
Version 2.1 November 2013 M. GIARDINO M.C. DE SANCTIS
(INAF-IAPS, (INAF-IAPS,
Rome) Rome)
M.T. CAPRIA
(INAF-IAPS,
Rome)
Version 2.4 May 2014 Chapter 8: added the M. GIARDINO M.C. DE SANCTIS
formula to calculate (INAF-IAPS, (INAF-IAPS, Rome)
the calibrated Rome) Rome)
reflectance factor M.T. CAPRIA
(INAF-IAPS, Rome)
Version 2.5 January 2016 Chapter 8: added the M. GIARDINO M.C. DE SANCTIS
name of the PDS field (INAF-IAPS, (INAF-IAPS, Rome)
containing the Rome) M.T. CAPRIA
integration time; (INAF-IAPS, Rome)
added information
about the ITF file
format.
Version 2.6 October 2016 Chapter 4: changed M. GIARDINO M.C. DE SANCTIS
denomination from (INAF-IAPS, (INAF-IAPS, Rome)
spectral shift to Rome)
spatial shift
Chapetr 8: added
description of the
detilt algorithm for
the visible channel
Version 2.7 February 2017 Chapter 8: added E. AMMANNITO M.C. DE SANCTIS
explanataion about the M. GIARDINO (INAF-IAPS, Rome)
parameters used in the (INAF-IAPS,
detilt algorithm; Rome)
updated the spectral
and temporal range
affected by the ITF
null values
Version 2.8 September 2017 Chapter 7: added E. AMMANNITO M.C. DE SANCTIS
description of internal M. GIARDINO (INAF-IAPS, Rome)
calibration procedure. (INAF-IAPS,
Created the new chapter Rome)
8 containing the
description of the
calibration refinement
obtained with the last
ITF. Previous chapter 8
("how to calibrate vir
in-flight data") has
been changed to chapter
9, while previous
chapter 9("references")
has been changed to
chapter 11.
Chapter 9: added an
explanation about the
presence of
contamination across
campaigns. Created
chapter 10 containing
the listing of knwon
instrument artifacts and
their effects on data
usability
Chapter 11: added a
new reference
Version 2.9 December 2017 General revision E. AMMANNITO M.C. DE SANCTIS
applied, during the PDS M. GIARDINO (INAF-IAPS, Rome)
peer review performed (INAF-IAPS,
on the Ceres volumes Rome)
between July and
December 2017
The internal
calibration steps have
been further documented
by inserting a table in
Chapter 7.
In Chapter 9, the
change operated to the
execution cycle of the
instrument to avoid the
contamination found in
VSH and VH2 is
described.
===========================================================================
This document describes the algorithms used to calibrate VIR raw (EDR) data to data in physical units (RDR, spectral radiance), in order to give the final user a detailed view of the methods used to remove instrumental effects on the data. A complete calibration campaign of VIR was performed at the channel level in Selex Galileo (SG), Florence by using a variety of calibration facility setups. These calibrations were performed immediately after instrument assembly and before delivery to Orbital for integration with the Dawn spacecraft. The SG calibration included spectral, geometrical, flat-field and radiometric measurements as well as characterization of the detectors' performances (including defective pixels, linearity and dark current at various operative temperatures), the instrumental thermomechanical stability, the data-commanding-telemetry handling, and electrical interfaces. A description of the methods used and results of these tests are described in De Sanctis et al. (2011).
Section 2 describes the experimental setup used for pre-launch calibrations at SG; section 3 is devoted to the description of the spectral calibration; geometrical calibration is included in section 4; flat-field is described in section 5; section 6 contains a description of radiometric calibration; section 7 is about the internal calibration procedure; section 8 describes an in-flight instrument response function update using the onboard lamp to improve the calibration in the spectral range 2.5 µm - 3.5 µm; and finally section 9 explains the algorithms used to convert raw data into spectral radiance or reflectance.
The basic setup used during the calibrations consists of an optical bench over which are housed a collimator, a reference target placed at its focal plane and a folding mirror used to move the collimated beam in the instrumental FOV along the azimuthal (sample) and zenithal (line) directions. Since VIR focuses at infinite distance it becomes necessary to use a collimator to have a collimated reference beam impinging the optical pupil. The SG-developed collimator uses an off-axis parabola (D=250 mm, F=1020 mm, off axis angle=8°), which guarantees an unobstructed beam, reduced aberrations, and high spatial scale. For VIR the magnification ratio is equal to:
MR=F_VIR/F_collimator=152mm/1020mm=0.15
This means that 1 mm on the collimator's focal plane corresponds to 0.15 mm on the VIR focal plane. The VIR detector has a 40 µm pixel pitch (square), so this scale corresponds to 4 spatial pixels along both sample and line directions. The collimator's focal plane is equipped with a holder able to sustain several interchangeable targets (pinholes, test slits, MTF masks, and a matrix of 5x5 microlamps). These elements are used to perform the different calibrations. The collimated beam is folded towards the instrument by using a folding mirror placed over two computer controlled, micrometric mounts able to aim it with steps of 1 µrad along the azimuthal (scan parallel to VIR slit, along sample direction) and zenithal (scan perpendicular to the slit, along lines direction) angles.
During the calibration the VIR instrument is housed in a thermo-vacuum chamber in order to reproduce the operating conditions aboard the satellite. In these conditions it's possible to cool the IR detector down to the operating temperature of about 70 K by using the cryo-cooler (operating on a Stirling cycle) and the CCD to about 160 K by using a passive radiator. The collimated optical beam reaches the spectrometer's pupil through a CaF2 window housed in the front of the thermo-vacuum chamber. This window is characterized by an elevated optical transmittance in the 250-5100 nm spectral range.
All opto-mechanical devices placed on the optical bench are controlled by using a dedicated software (OCS, Optical Control System), while the VIR instrument is controlled by using a separate setup, consisting of the UT (Unit Tester) connected to the experiment through the Proximity Electronics Module (PEM). This system allows the sending of commands to the instrument, to start acquisitions only when all optical elements commanded by OCS are in the correct configuration, and to receive back and record telemetry and calibration data.
The spectral calibration characterizes a fundamental aspect of the functional requirements of a hyperspectral imaging spectrometer: the conversion of bands positions along the spectral axis of the detectors into wavelength units. The spectral calibration is obtained through the following steps:
The following instrumental parameters are deduced from the spectral calibration:
Because the instrument uses a diffraction grating that disperses the light according to a linear law we can assume SSI(n) = SSI; in this case the spectral calibration relation assumes the following expression for the spectral dispersion of the central wavelength:
λc(n) = λ0 + SSI • b
These quantities were measured during the ground calibration by acquiring several fine spectral scans using a monochromator as a source. The calibration setups used to define the spectral properties of VIR used a heritage setup developed for the VIRTIS/M aboard Rosetta and the Venus Express missions (Ammannito et al. 2006; Filacchione et al. 2006). Two different configurations were used to characterize the spectral response, the first using a transmission method and the second using a diffusion method. In the first case (transmission) on the optical bench, the source, the monochromator, the test slit and the collimator were present; using this set-up the level of the signal was high enough to stimulate VIR, but the alignment between the output slit of the monochromator and the test slit of the optical bench was difficult to achieve. In the second case (diffusion) the source, the monochromator, a silvered diffusive target and the collimator were on the bench. In this case the alignment of the system wasn't critical, but the level of the signal was lower. The monochromator scans different wavelengths, thus illuminating the diffusive screen. VIR acquires monochromatic images at each step. Therefore VIR is "simulated" at wavelength steps smaller than its spectral resolution. In this way it is possible to associate to each frame the wavelength of the input beam coming from the monochromator. Studying the profile over the lines of each illuminated band, it is possible to get the spectral response function of that particular band.
The transmission method was applied only to the visual channel. For each illuminated pixel, the spectral response function was computed by fitting a Gaussian-like function over the profiles measured during the spectral scans. With this setup the intensity of the signal along the slit isn't uniform. This could be related to a misalignment between the output slit of the monochromator, the test slit of the optical bench in particular, and the entrance slit of the experiment. Moreover on the optical bench a spectral shift along the slit is apparent, given that at every illuminated band a different central wavelength is found. The central wavelength and the spectral width of the illuminated bands are calculated by averaging such parameters over every illuminated sample. The central wavelength of the band is calculated by using a linear fit while the spectral width is given by a polynomial fit. A summary of the parameters calculated with the fits and their uncertainties are reported in the following table:
Spectral dispersion Spectral width
Model a • x + b a • x4 + b • x3 + c • x2 + d • x + e
Parameters a=1.89297 b=245.744 a=5.25E–11 b=-6.08E–8 c=2.74E–5 d=-0.0049 e=2.13
Sigma σa=0.00016 σb=0.041 σa=0.14E–11 σb=0.14E–7
σc=4.82E–5 σd=0.0068 σe=0.31
In the next table are indicated, for all the illuminated bands, the measured and calculated values of the central wavelength and spectral width. The measured values are averages computed over all the illuminated samples for a fixed band.
Band # λmeas λcal SWmeas SWcal
(nm) (nm) (nm) (nm)
79 395.125 395.289 1.89559 1.88451
80 397.049 397.182 1.89332 1.88290
81 398.964 399.075 1.88235 1.88131
82 400.860 400.967 1.87189 1.87976
83 402.761 402.860 1.86246 1.87824
84 404.662 404.753 1.87010 1.87675
157 542.828 542.94 1.87158 1.82947
158 544.925 544.833 1.84376 1.82947
159 546.835 546.726 1.83146 1.82949
160 548.742 548.619 1.82906 1.82952
161 550.640 550.512 1.82154 1.82956
162 552.548 552.405 1.84028 1.82961
163 554.454 554.298 1.84128 1.82967
237 694.480 694.378 1.81275 1.85861
238 696.323 696.271 1.86955 1.85922
239 698.219 698.164 1.84252 1.85983
240 700.101 700.057 1.85477 1.86045
241 702.000 701.950 1.86519 1.86107
242 703.898 703.843 1.85821 1.86170
317 845.842 845.816 1.91248 1.91559
318 847.74 847.709 1.91959 1.91639
319 849.647 849.602 1.92232 1.91720
320 851.542 851.495 1.91718 1.91800
321 853.447 853.388 1.91865 1.91881
396 995.282 995.361 1.97795 1.99122
397 997.171 997.254 1.98837 1.99243
398 999.063 999.147 1.99155 1.99365
399 1000.95 1001.04 1.98800 1.99487
400 1002.85 1002.93 2.00388 1.99611
401 1004.75 1004.83 2.00954 1.99735
From the analysis of measurements we have noted the presence of a slight spectral shift occurring along the slit. In order to evaluate this effect we repeated the calculation of the linear fit coefficients for other samples along the slit (at samples = 110, 140); previously, the results at the slit's center (sample = 128) were discussed. This analysis demonstrates the presence of a change in the spectral calibration response when repeated on different points along the slit. The fit parameters with their errors on samples 110, 128, 140 are reported in the following table.
Sample# λmeas SWmeas
(nm) (nm)
110 246.76±0.56 1.8926±0.0022
128 245.83±0.30 1.8926±0.0011
140 245.40±0.19 1.8921±0.0007
These calculations demonstrate that the parameters are mutually incompatible so the central wavelengths calculated with the coefficients indicated in the previous table cannot be used and another calibration approach is necessary. At the end of the post-processing analysis it was concluded that the measurements of the spectral width are satisfactory while further measurements are needed to determine the dispersion coefficients. Using the Transmission setup, in fact, they seem to be sample dependent. In the next paragraph the results obtained using the Diffusion setup data are discussed.
The Diffusion method was used to characterize the spectral response of both the Visual and the Infrared channels. We use these measurements to determine the central wavelength of the VIS channel and both the central wavelength and the spectral width of the Infrared channel. Comparing these results with the similar profiles taken with the Transmission setup, it is apparent that the spectral shift among profiles taken at different samples is negligible. In this way it is verified that the effect is caused by the Transmission set-up characteristics (difficult to co-align VIR and test slit orientations) and is not due to the VIR malfunctioning.
Using the same technique discussed previously, we have retrieved the best spectral dispersion and width values. For both channels, the central wavelength of each band b is retrieved through a linear fit while for the spectral width a polynomial fit is used.
Spectral dispersion Spectral width
VIS Model a • x + b a • x4 + b • x3 + c • x2 + d • x + e
Parameters a=1.89223 b=245.660 a=1.3E–10 b=-1.1E–7 c=1.89E–5 d=0.0047 e=1.6
Sigma σa=0.00033 σb=0.085 σa=8.1E–10 σb=7.8E–7
σc=0.26–5 σd=0.0037 σe=1.7
IR Model a • x + b a• x4 + b • x3 + c • x2 + d • x + e
Parameters a=9.4593 b=1011.29 a=-6.8E–10 b=8.23E–7 c=-2.09E–4 d=0.0021 e=13.9
Sigma σa=0.0011 σb=0.28 σa=2.3E–9 σb=1.6E–6
σc=3.80E–4 σd=0.0335 σe=1.2
We report in the next Table the measured and computed values of the central wavelength and spectral width for both channels. The measured values are averaged over all the samples for a given band.
VIS channel
Band λmeas λcal SWmeas SWcal
# (nm) (nm) (nm) (nm)
81 398.875 398.931 2.07338 2.03109
82 400.774 400.823 2.02554 2.03699
83 402.597 402.715 1.99148 2.04287
84 404.436 404.607 2.07068 2.04875
158 544.700 544.633 2.42099 2.44303
159 546.577 546.525 2.44812 2.44754
160 548.465 548.417 2.45310 2.45203
161 550.355 550.309 2.45829 2.45648
162 552.249 552.201 2.47181 2.46091
163 554.144 554.094 2.47652 2.46531
164 556.028 555.986 2.46822 2.46969
238 696.007 696.011 2.69859 2.70952
239 697.892 697.903 2.69611 2.7116
240 699.785 699.795 2.69541 2.71365
241 701.677 701.688 2.70452 2.71567
242 703.572 703.58 2.72082 2.71767
243 705.508 705.472 2.80885 2.71963
317 845.453 845.497 2.75994 2.79344
318 847.337 847.389 2.80736 2.79368
319 849.239 849.282 2.8061 2.79391
320 851.142 851.174 2.79165 2.79412
321 853.043 853.066 2.80487 2.79433
396 994.789 994.983 2.75589 2.80388
397 996.686 996.875 2.8257 2.80439
398 998.595 998.768 2.77639 2.80492
399 1000.46 1000.66 2.79735 2.80548
400 1002.28 1002.55 2.78868 2.80606
401 1004.22 1004.44 2.84745 2.80666
IR channel
Band λmeas λcal SWmeas SWcal
# (nm) (nm) (nm) (nm)
2 1029.3 1030.21 14.0742 13.9467
3 1038.77 1039.67 13.78 13.9478
103 1986.31 1985.6 12.9869 12.7654
104 1995.85 1995.06 12.7585 12.7477
105 2005.35 2004.52 12.6494 12.7299
106 2014.87 2013.98 12.5963 12.7122
208 2978.82 2978.83 11.4667 11.4689
209 2988.07 2988.29 11.2923 11.4666
210 2997.45 2997.75 11.6996 11.4645
211 3006.83 3007.21 11.5063 11.4626
212 3016.13 3016.67 11.337 11.461
315 3991.6 3990.98 12.7906 12.9028
316 4000.3 4000.44 12.6859 12.9335
317 4010.2 4009.9 13.5031 12.9647
367 4482.68 4482.86 14.9175 14.9333
368 4492.2 4492.32 15.1484 14.9807
369 4501.56 4501.78 14.8612 15.0283
370 4511.02 4511.24 15.1136 15.0763
Following the post-processing analysis, we find that the measurements of the dispersion coefficients are compliant with the specifications for both the visual and infrared focal planes. For the VIS channel, computation with the diffusion method gives better results than computation with the transmission method. The quality of the spectral calibration was confirmed by observing the spectrum of a calibrated HgNe pencil lamp.
The geometrical calibration allows characterization of:
1. the field of view, hereafter FOV;
2. the instantaneous field of view (hereafter IFOV) of different pixels along and across the spectrometer's slit directions (respectively sample and line directions).
We define the pixel function, PF(s), as the convolution of a unitary step function V (s) (representing the real pixel) with the instrumental response along the sample direction, INST(s):
PF(s) = V(s) ⊗ INST(s)
The slit function, SF(l), is given by the convolution of a unitary step function U(l) (representing the spectrometer's slit response) with the telescope response along the line direction, TEL(l):
SF(l) = U(l) ⊗ TEL(l)
These two responses were measured during the pre-launch calibration campaign acquiring the signal produced by a test-slit, illuminated by a HgNe lamp, having an equivalent width narrower than the instrumental IFOV (the test slit aperture is 3.0 x 0.1 mm, corresponding to 12 x 0.4 pixels at VIR scale). This test-slit is placed at the collimator's focus and it is moved at subpixel steps perpendicular and parallel to the VIR slit by moving the folding mirror. By using this method it is possible to measure the FWHM of the IFOV at three positions of the FOV (boresight: sample = 128, line = 128), position N: sample = 38, line = 218; position O: sample = 218, line = 38). For the VIS channel the FWHM of the pixel function ranges over the 237.9–244.1 µrad interval while the slit function is 287.7–389.4 µrad; for the IR channel the ranges are 421.7–488.1 and 350.9–367.3 µrad respectively. These differences are caused by a residual of astigmatism in the optical design.
The determination of the FOV (nominally 3.6° x 3.6°) is possible through the imaging of a 5 x 5 array of microlamps placed at the focus of a collimator. This array was built to cover the entire FOV when placed at collimator's focus: the presence of a regular grid of subpixel sources allows for evaluation of the imaging and geometrical performances of the experiment. The absolute position of each microlamp was measured with a theodolite placed on the pupil of the collimated beam; when compared to the relative positions of the lamps spots on the images it is possible to infer the dimensions of the instrumental FOV.
Moreover, this setup is particularly useful in evaluating the presence of possible "spatial shift", e.g. a mismatch between the position of one monochromatic image with respect to another. This effect is particularly evident on VIRTIS-M on Rosetta, where it reaches a shift of about 8 spatial pixel (samples) between the first and the last image of the VIS channel. The cause of it is a slight misalignment among slit, grating grooves and focal plane orientation (for a full discussion of the spectral tilt and post-processing corrective methods the reader can refer to Filacchione 2006). For VIR several optical improvements were made to the grating design that drastically reduce this effect. Analysis of the 25 microlamp target data allows verification that the spatial shift on the VIS channel reaches about 2 spatial pixels between the two spectral extremes of the range ( 255 nm and 1071 nm ). This value comes from the analysis of the distribution of the microlamps' position (in sample-line space) on the monochromatic images. As each microlamp has a subpixel dimension when seen by VIR through the optical bench setup, it is possible to measure the associated barycenter position through a 2D Gaussian fit; this procedure is done for each lamp and for every spectral band (432 images).
The flat-field is defined as the response of the instrument to a uniform source (Filacchione et al. 2006). It is used to homogenize the pixels' response across the whole focal plane. In the case of imaging spectrometers using 2D detectors, flat field matrices contain, for each wavelength, the relative variation of the instrumental response with respect to the boresight (sample s* = 127).
The measurements of the VIS and IR flat-field matrices were calculated during the pre-launch tests by acquiring a spatially flat source placed on the focus of a collimator and aligned to the VIR boresight. The source used in the 0.25–2.5 µm range is a Lambertian surface illuminated by a QTH lamp; this target is about 10 x 10 cm wide in order to completely fill the instrumental FOV. It is replaced by a blackbody source for the measurement of the flat-field in the 2.5–5.0 µm range. In both cases the flat field is retrieved through a spatial scan across these targets by moving the folding mirror at 1 IFOV step. This approach allows for observation of the same region of the target with each pixel (sample) of the detector, thus eliminating possible target non-uniformity from the flat-field matrices.
The resulting flat-field matrices for the two focal planes are given by the ratio of the signal measured at a certain position of the focal plane (b, s) with respect to the signal measured at boresight (s = s*) and at the same band position b:
FF(b,s) = Ns(b,s) / Ns(b, s*)
Flat-field matrices are sensitive to the characteristics of the detector (single and clusters of defective pixels, dis-uniformities due to the production process) and of the optical layout (the two horizontal features at samples 80 and 150 are caused by the slit's shape; several vertical features with a symmetry with respect to boresight are introduced by the grating design).
As explained in section 5, the wide spectral range of the experiment can only be explored by using different sources (Filacchione et al., 2006). For the radiometric calibration two different sources are necessary:
The input radiance is measured and verified through a laboratory radiometer (a Field-Spec(TM) spectroradiometer). Unfortunately as the optical pupil of the Field-Spec optics does not match entirely with the VIR pupil, the measured radiance can only be used as a relative value. The absolute value of the radiance has been tested in flight during dedicated observations of known targets, such as stars (Arcturus and Canopus) and planets (Mars). The lamps used are observed first with the spectroradiometer and then with VIR. Knowing the value of input radiance, we can associate it with an average of 50 VIR acquisitions of the Spectralon(TM) target, taken at slit center, with an integration time ti = 10 s. The Responsivity, R, is therefore calculated by applying the following equation:
R(b, s*) = DN(b, s*) / (BB(b)*ti)
where R(b,s*) is the responsivity computed for each band b at the sample s*, DN(b,s*) is the raw signal in digital numbers acquired by the spectrometer for each band b at the sample s*, BB(b) is the radiance of the source measured by the reference spectroradiometer and sampled at the VIR spectral band b and ti is the integration time.
The expansion to the sample of the focal plane different from s* is possible applying the flat-field FF. In this way we retrieve the ITF (Instrument Transfer Function) array:
ITF(b, s) = FF(b, s) • R(b, s*)
The IR channel radiometric calibration is done by acquiring directly the radiance emitted by a blackbody source placed at the collimator's focus. The blackbody temperature is set at different values in order to have a good SNR on several spectral ranges and with different integration times (a summary of the acquisitions is given below). As reported in the next Table, only a limited spectral range can be evaluated for a given blackbody temperature and integration time: for bands < Min Band the signal is very low and it includes only the readout offset and residuals of the dark current; for bands > Max Band value the signal is saturated. Therefore, the responsivity is retrieved by using only the signal intervals as indicated in following:
TBB (°C) ti (s) Min Band Max Band
50 0.2 250 438
1.0 238 280
2.0 238 255
5.0 170 240
100 0.2 238 281
1.0 148 239
2.0 140 195
5.0 120 170
200 0.2 110 174
1.0 80 120
2.0 70 105
5.0 65 95
300 0.2 60 100
1.0 40 68
2.0 35 58
5.0 0 37
350 0.2 0 78
1.0 0 52
2.0 0 35
The IR responsivity is computed by using:
R(b, s*) = DN(b, s*) / (BB(b) * ti)
where the blackbody radiance BB is given by Planck's formula. Finally, applying
ITF(b, s) = FF(b, s) • R(b, s*)
to these data, it is possible to derive the responsivity for each pixel of the IR channel.
Instrumental performances were checked during in-flight conditions by using internal calibration sequences. VIR can acquire reference signals by using the combination of the cover, shutter and VIS and IR lamps (Melchiorri et al., 2003). These lamps, housed on the side of the telescope illuminate the internal side of the external cover. The cover is placed near the entrance pupil of the instrument to minimize optical aberrations. The window of each lamp contains a transparent filter (holmium for the VIS, polystyrene for the IR) to introduce some well-shaped spectral absorption features on the overall spectrum. The signal coming from the two lamps can be used to:
- check the in-flight stability of the instrument spectral response;
- check the in-flight stability of the flat-field;
- monitor the evolution of defective pixels (number and distribution);
- perform a check on the relative radiometric response of the instrument.
The internal calibration mode, implemented in the VIR on-board software, consists of the acquisition of a sequence of 35 frames organized into the following steps: 5 acquisitions of electronic offsets, 5 acquisitions of background signal, 5 acquisitions of dark currents, 5 acquisitions of the IR lamp signal, 5 acquisitions of the VIS lamp signal, 5 more acquisitions of dark currents and finally 5 acquisitions of background signal.
The electronic offset is measured by acquiring 5 frames with both detectors on with an exposure time of zero seconds, while maintaining the instrument cover closed and the shutter open, so to read only the "electronic noise produced inside" the instrument chassis, isolated from any signal coming from outside.
The background signal is instead measured by acquiring 5 frames in the same condition defined for the electronic offset, except for setting an exposure time greater than zero for both detectors.
The dark current frames are produced while keeping both the shutter and the cover closed and activating both detectors.
The measurements of the dark currents and that of the background signal are repeated twice during calibration mode, to record their values both before and after the activation of the internal calibration lamps. Finally, the signal from the calibration lamps is measured by activating the given calibration lamp (IR or VIS), with the shutter open and the cover closed with the internal side illuminated by the lamp itself.
Calibration is done illuminating with one lamp at a time.
In the following table these different phases are listed.
===========================================================================
Phase Frames Cover Shutter IR VIS Exp. Time Exp Time
Number lamp lamp VIS(s) IR(s)
===========================================================================
Electronic 1-5 closed open off off 0.0 0.0
Offset
Background 6-10 closed open off off 1.0 0.5
1
Dark 11-15 closed closed off off 1.0 0.5
Current 1
IR Lamp 16-20 closed open on off 20.0 0.5
VIS Lamp 21-25 closed open off on 1.0 0.02
Dark 26-30 closed closed off off 1.0 0.5
Current 2
Background 31-35 closed open off off 5.0 5.0
2
===========================================================================
[INTERNAL CALIBRATION SEQUENCE]
Data acquired during this sequence are fundamental to follow the temporal evolution of the instrument and to monitor the overall performances in operative conditions. They can also be used in the calibration pipeline.
During the mission operations at Vesta, we identified some artifacts in the instrumental transfer function (ITF) in the 2.5–3.5 µm spectral range, where several absorption bands of OH and H2O occur. These artifacts are systematic and of the same relative magnitude in all images, and therefore they do not prevent the detection of relative spectral variations associated with OH and H2O on the surfaces of the target bodies. These artifacts are systematic errors due to the non-homogenous instrumental response of a detector producing nonphysical spectral signatures. The precise cause of the artifacts are not yet fully understood, but involve separately or in combination imperfect radiance calibration, spectral miscalibration, peculiar readout noise in the detector electronics, and from uncertainties in the solar reference spectrum. Nevertheless, we have devised a means to correct the ITF using the onboard lamps.
To compute the ITF for this range, we used in-flight data from one of the internal lamps of the spectrometer. This lamp, made of a tungsten filament, is characterized by a blackbody-like emission at about 2400 K. Since the spectrum of the infrared radiation emitted by these filaments is featureless, a polystyrene filter was inserted for a wavelength calibration of the IR channel. The blackbody radiation of the internal lamp has been used to retrieve a relative ITF in the 2.5–3.5 µm spectral range. First, we calibrated the signal from the internal calibration lamp with the on-ground response function, and then we retrieved the equivalent temperature of the radiation fitting a Planck curve, a value around 1500 Kelvin. The new ITF is the result of the ratio between the raw signal of the lamp and the Planck function. This ratio must be multiplied by a factor to take into account the integration time used to acquire the signal, the transmittance of the polystyrene filter, and the viewing geometry. We compared the ITF obtained with the on-ground calibration with the in-flight calibration and a combination of the two, and we analyzed the calibrated spectra computed with these three versions of the ITF: the comparison graphs between these three curves can be found in [8]. We observe that the new ITF minimizes most calibration residuals that were showing as artifact peaks between 2.5 and 2.9 µm in the previous calibration. The final ITF (version 2) is the ITF derived during the on-ground calibration campaign with the exception of the spectral channels between 2.5 and 3.5 µm where the ITF is derived using the method described here.
The VIR team receives data and telemetry packets from the satellite from the Dawn Science Center (UCLA-JPL). These packets are processed at the PI institution (INAF-IFSI, Rome, Italy) with proprietary GSE (Ground Support Equipment) and converted into standard PDS (Planetary Data System) format. Dedicated package scripts and routines and calibration files distributed with this archive are used to convert the raw data in physical units.
Only for the visible channel, a specific detilt algorithm must be applied to the raw cube to deal with the spatial shift described in chapter 4. This algorithm is applied as a first step in the calibration procedure, before any other step described in this section.
The spatial tilt effect in the visible channel can be compensated by a shift of equal magnitude directed towards the opposite axis of the sample.
The detilting is obtained through the following steps:
The value 20 as a multiplicative factor, has been chosen to find a suitable compromise between the accuracy of the algorithm and the computational time required to obtain the result.
The spatial offset applied is a constant value of 2 samples along the slit. As a result, the frame obtained has the same size of the original frame, but the two ending columns (corresponding to two samples) are filled with empty values and are unusable.
The routine implementing the steps above is shown in the following IDL code snippet: the input data for this routine is the raw cube (raw_qube), while the output data is the detilted raw qube(raw_qube_detilt).
frame_expanded=uintarr(bands,samples* 44)
frame_exp_detilt=uintarr(bands,samples * 40)
raw_qube_detilt=uintarr(bands,samples,lines)
; number of pixel by which each sample in each spectral channel has to be shifted
for li=0,lines-1 do begin
; step 1: oversampling
for sa=0,samples-1 do begin
for ss=0,39 do begin
frame_expanded(*,sa*40+ss)=reform(raw_qube(*,sa,li))
endfor
endfor
; step2: detilt
for sa=0,samples-1 do begin
for ba=0,bands-1 do begin
bsh=ba/4
for ss=0,39 do begin
frame_exp_detilt(ba,sa*39+ss)=frame_expanded(ba,sa*39+ss+bsh)
endfor
endfor
endfor
; step3: resampling to the original size
for sa=0,samples-1 do begin
tot=dblarr(bands)
for ba=0,bands-1 do begin
for ss=0,39 do begin
tot(ba)=frame_exp_detilt(ba,sa*40+ss)+tot(ba)
endfor
raw_qube_detilt(ba,sa,li)=tot(ba)/40.0
endfor
endfor
endfor
The Software Interface Specification (SIS) document contains the details of VIR data cube format and processing steps, detailed briefly here. A raw data cube contains uncalibrated signal Ns in DN; dark currents are periodically stored in the same raw data cube and in each data cube there is at least one dark current acquisition. The dark current must be subtracted from the original data in the raw cube before the conversion in physical units. The number and location of dark current frames in each raw cube is documented in the hkt table (shutter status, open if normal acquisition, closed if dark current acquisition). The same information alternatively can be found by reading the parameter DARK_ACQUISITION_RATE in the data cube label file.
Raw data cubes may have one or more dark current frames. If there is only one dark current frame in a raw data cube, the equivalent dark current frame is the same for every frame in the data cube and is equal to the only dark current frame acquired, If there is more than one dark current frame, the equivalent dark current frame is the interpolation in time of two consecutively acquired dark frames. The dark subtracted frames are computed subtracting the equivalent dark current frame from the original frame in the raw data cube. At the end of this operation the dark current frames are removed, and there will be a dark subtracted data cube with the same bands and samples number of the raw data cube and a number of lines equal to the original minus the number of dark current frames. This is the reason why calibrated cubes have a lower number of frames than the corresponding raw cubes.
The counts stored in the PDS cube can be converted into physical units of spectral radiance Rad (W m-2 µm-1 sterad-1) by using the following equation:
S(λ(b), s, l) = Ns(b, s, l) / (ITF( λ(b), s)* ti)
where:
-S(λ(b),s,l) is the cube calibrated in spectral radiance which have the same bands and samples number of the raw data cube and a number of lines equal to the original minus the number of dark current frames of the raw cube;
- λ(b) is the wavelength associated to band b according to spectral calibration tables of VIS and IR channels (files DAWN_VIR_VIS_HIGHRES_SPECAL_Vx.TAB and DAWN_VIR_IR_HIGHRES_SPECAL_Vx.TAB, respectively);
-s, l corresponds to sample and line location of the pixel in the dark subtracted cube;
-ti is the integration time of the observations (in seconds) as indicated in the field FRAME_PARAMETER .EXPOSURE_DURATION of PDS header of the file for VIS and IR channels;
-ITF(λ(b), s) is the Instrument Transfer Function matrix for VIS and IR channels (files DAWN_VIR_VIS_RESP_Vx.DAT and DAWN_VIR_IR_RESP_Vx.DAT, respectively).
At the same time, to calculate the calibrated reflectance factor (sometimes termed 'I/F'), the equation is:
R(λ(b), s, l) = (S(λ(b), s, l) * ( π * (ssd / K) 2)) / si
where
-R(λ(b), s, l) is the cube calibrated reflectance factor which has the same number of bands and samples as the raw data cube and a number of lines equal to the original minus the number of dark current frames of the raw cube;
- K is the value of one astronomical unit expressed in km ( 149597870.7 ) ;
-ssd is the spacecraft heliocentric distance expressed in km, as read from the cube label file in the SPACECRAFT_SOLAR_DISTANCE field;
- si is the solar irradiance for VIS and IR channels (files DAWN_VIR_VIS_SOLAR_SPECTRUM_Vx.DAT and DAWN_VIR_IR_SOLAR_SPECTRUM_Vx.DAT, respectively).
These calculations can be applied to high resolution acquisitions (432 bands times 256 samples); in nominal modes, where spatial and/or spectral resolutions are reduced, it is necessary to interpolate both spectral tables and responsivity matrices according to binning values.
The following calibration files are stored in the CALIB directory of the PDS archives:
- DAWN_VIR_VIS_RESP_Vx.DAT, a 432x256 floating precision matrix containing the VIR-VIS Instrumental Transfer Function, including the VIS flat-Field. The file format is binary, matrix values are stored using double precision floating point precision, band interleaved (PDS type is IEEE_REAL with 8 bytes length)
- DAWN_VIR_IR_RESP_Vx.DAT, 432x256 floating precision matrix containing the VIR-IR Instrumental Transfer Function, including the IR flat-Field.
- DAWN_VIR_VIS_HIGHRES_SPECAL_Vx.TAB and
- DAWN_VIR_IR_HIGHRES_SPECAL_Vx.TAB, 432 row ASCII tables containing the wavelengths of the VIS and IR channels in High Resolution Mode.
- DAWN_VIR_VIS_WIDTH432_Vx.TAB and
- DAWN_VIR_IR_WIDTH432_Vx.TAB, 432 row ASCII tables containing the width of the VIS and IR channels in High Resolution Mode.
These files must be used for cubes collected in High Resolution Mode.
Cubes in Nominal Mode (x3 binning along bands) can be calibrated by using the following spectral calibration files:
- DAWN_VIR_VIS_NOMRES_SPECAL_Vx.TAB and
- DAWN_VIR_IR_NOMRES_SPECAL_Vx.TAB, 144 row ASCII tables containing the wavelengths of the VIS and IR channels in Low Resolution Mode.
- DAWN_VIR_VIS_WIDTH144_Vx.TAB and
- DAWN_VIR_IR_WIDTH144_Vx.TAB, 144 row ASCII tables containing the width of the VIS and IR channels in Low Resolution Mode ("x" is a digit representing the version number of the file). The first release is "V1".
VIR data included in this release are calibrated by using this basic pipeline. Further improvements, based on the use of the internal calibration sequences, may be included in future data releases. The ITF is also currently under improvement. Calibrated values in the spectral range [2.818µm - 3.272µm] are still under verification. These values have been put to null in the ITF only for the acquisitions performed during the mission campaigns VSH and VH2; during these two campaigns, a major external contamination is taken as responsible for the artifacts affecting the spectral range around 3 micrometers. The effects of the contamination on this spectral range have been avoided for acquisitions performed after these campaigns. This result was obtained by commanding the instrument into a given operative mode characterized during the analysis of this phenomenon. Such analysis come to the decision of minimizing the duration of the cryo-cooler activity, switching it off whenever compatible with the planned observation. Specifically, the cryo-cooler was always turned off at the end of each observation sequence.
The team is also preparing a specific algorithm to apply a correction for these two campaigns that will be released in the future.
VIR spectra are affected by residual systematic errors due to imperfect radiometric and spectral calibration that influence the quality of imaging spectrometer data. These include systematic deviations from spectrum of the target due to imperfect radiometric standards, by spectral miscalibration, and by systematic errors resulting from uncertainties in the solar reference spectrum.
The spectral images can show 'stripes' due to the slight deviation that exists between the input/output transfer function of each sample of the detector. These stripes are particularly evident when the signal is very low.
Visible channels with a wavelength > 0.95 microns cannot be used for scientific analysis, because of the straylight effect which currently has no correction. By excluding these wavelengths from the spectrum, the offset between the visible and the infrared channels in the range where they meet disappears. In some sporadic cases where this offset can still be observed, the instrument team recommends scaling the IR channel to the VIS channel, as the latter is in a good agreement with the Framing Camera dataset.
Defective pixel and filters boundaries should be not considered for any scientific analysis. The filter boundaries and defective pixels are listed in the tables below. Bands contiguous to the filter boundaries may be affected by straylight: the presence and intensity of straylight depends on the specific conditions at which each spectrum has been acquired. Therefore, the user shall check the presence of this effect and eventually discard the bands affected.
A discontinuity in the band pass filters between filters 3 and 4 results in a discontinuity in the transmissivity near 2.4 microns. When the data are corrected for variations in transmissivity a spike is introduced at this discontinuity (see fig. 8b in [3]).
VIR spectra are affected by a positive slope in the VIS-NIR range when compared to ground based spectra of the same target (Vesta and Ceres). Although the origin of this effect is not currently understood, we decided to re-normalize the VIR dataset to correct this effect. The correction is a scale factor computed as the ratio between a ground based reference and VIR spectrum of Ceres. For further details see [8].
The instrument team is working to resolve some of the issues described here. The team plans to release, in the near future, a new level of derived data products that will partially solve the problems above reported. The 'artifact removed' data set is planned for future release.
Dropouts that may be observed in some images are due to instrument saturation. The spectral range in which these dropouts occur varies depending upon exposure duration, target topography, and incidence angle.
In the following table, the filter boundaries positions are listed.
===========================================================================
CHANNEL SAMPLE INTERVAL BAND INTERVAL WAVELENGTH INTERVAL (nm)
VIS 1 - 256 222-223 673.30398 - 675.19621
IR 1 - 256 49 - 54 1474.79668 - 1522.09328
IR 1 - 256 156 - 161 2486.94392 - 2534.24052
IR 1 - 256 290 - 293 3754.4928 - 3782.87076
IR 1 - 256 357-360 4388.26724 - 4416.6452
===========================================================================
[ VIR DETECTOR FILTERS BOUNDARIES]
The defective pixels are listed below.
===========================================================================
CHANNEL SAMPLE BAND INTERVAL WAVELENGTH INTERVAL (nm)
VIS 30 308 836.03576
VIS 31 308 836.03576
VIS 47 409 1027.15099
VIS 48 187-188 607.07593-608.96816
VIS 49 59 364.87049
VIS 54 137 512.46443
VIS 71 215 660.05837
VIS 100 78 400.82286
VIS 108 413 1034.71991
VIS 109 19 289.18129
VIS 111 19 289.18129
VIS 114 424 1055.53444
VIS 118 363 940.10841
VIS 126 410 1029.04322
VIS 130 292 805.76008
VIS 136 271 766.02325
VIS 139 235 697.90297
VIS 147 222 673.30398
VIS 150 54 355.40934
VIS 150 59 364.87049
VIS 150 78 400.82286
VIS 160 372 957.13848
VIS 162 36-37 321.3492-323.24143
VIS 162 248 722.50196
VIS 162 330 877.66482
VIS 163 36-37 321.3492-323.24143
VIS 163 248 722.50196
VIS 163 330 877.66482
VIS 165 32 313.78028
VIS 166 32 313.78028
VIS 166 173 580.58471
VIS 168 232 692.22628
VIS 169 363 940.10841
VIS 172 189 610.86039
VIS 173 92 427.31408
VIS 175 228 684.65736
VIS 175 266-267 756.5621-758.45433
VIS 176 152 540.84788
VIS 176 229 686.54959
VIS 177 155 546.52457
VIS 179 196 624.106
VIS 181 249 724.39419
VIS 183 354 923.07834
VIS 186 238 703.57966
VIS 186 387 985.52193
VIS 188 276 775.4844
VIS 188 352 919.29388
VIS 189 294 809.54454
VIS 189 352 919.29388
VIS 189 391 993.09085
VIS 189 413 1034.71991
VIS 190 195 622.21377
VIS 191 411 1030.93545
VIS 194 358 930.64726
VIS 196 266 756.5621
VIS 196 362 938.21618
VIS 199 23-24 296.75021-298.64244
VIS 203 257 739.53203
VIS 203 370 953.35402
VIS 204 257 739.53203
VIS 207 265 754.66987
VIS 211 291 803.86785
VIS 216 287 796.29893
VIS 222 249 724.39419
VIS 222 338 892.80266
VIS 223 339-340 894.69489-896.58712
VIS 225 274 771.69994
VIS 227 103 448.12861
VIS 229 248 722.50196
VIS 234 306 832.2513
VIS 234 424 1055.53444
VIS 238 249 724.39419
VIS 238 277 777.37663
VIS 238 416-417 1040.3966-1042.28883
VIS 239 405 1019.58207
VIS 241 15-16 281.61237-283.5046
VIS 241 386-387 983.6297-985.52193
VIS 242 15-16 281.61237-283.5046
VIS 242 364 942.00064
VIS 245 128 495.43436
VIS 248 304-305 828.46684-830.35907
VIS 250 223 675.19621
VIS 251 223 675.19621
VIS 252 274 771.69994
VIS 253 307 834.14353
IR 8 86 1824.79152
IR 12 148 2411.26936
IR 16 327 4104.48764
IR 20 39-43 1380.20348-1418.04076
IR 21 39-42 1380.20348-1408.58144
IR 22 40-42 1389.6628-1408.58144
IR 27 374 4549.07568
IR 35 218 3073.42176
IR 45 337 4199.08084
IR 51 212 3016.66584
IR 52 280 3659.8996
IR 56 430 5078.7976
IR 74 121 2155.86772
IR 79 185 2761.2642
IR 79 190 2808.5608
IR 82 190 2808.5608
IR 84 188 2789.64216
IR 86 182 2732.88624
IR 86 200 2903.154
IR 92 30 1295.0696
IR 94 189 2799.10148
IR 99 73 1701.82036
IR 100 73 1701.82036
IR 101 223-224 3120.71836-3130.17768
IR 102 72 1692.36104
IR 102 223 3120.71836
IR 102 225 3139.637
IR 103 223 3120.71836
IR 111 304 3886.92328
IR 112 28 1276.15096
IR 121 193 2836.93876
IR 122 172 2638.29304
IR 128 149 2420.72868
IR 128 187 2780.18284
IR 130 195 2855.8574
IR 132 182 2732.88624
IR 136 344 4265.29608
IR 138 383-384 4634.20956-4643.66888
IR 140 202 2922.07264
IR 142 341-342 4236.91812-4246.37744
IR 143 343 4255.83676
IR 144 343 4255.83676
IR 145 343 4255.83676
IR 146 342 4246.37744
IR 146 344 4265.29608
IR 148 108 2032.89656
IR 149 169-170 2609.91508-2619.3744
IR 155 1 1020.74932
IR 156 1-9 1020.74932-1096.42388
IR 156 196 2865.31672
IR 157 1-15 1020.74932-1153.1798
IR 157 25 1247.773
IR 158 9-17 1096.42388-1172.09844
IR 159 14-18 1143.72048-1181.55776
IR 160 19-20 1191.01708-1200.4764
IR 160 28-29 1276.15096-1285.61028
IR 161 26 1257.23232
IR 161 28-29 1276.15096-1285.61028
IR 161 181 2723.42692
IR 171 57-64 1550.47124-1616.68648
IR 172 57-64 1550.47124-1616.68648
IR 172 227 3158.55564
IR 173 59-68 1569.38988-1654.52376
IR 174 60-67 1578.8492-1645.06444
IR 175 61-63 1588.30852-1607.22716
IR 191 111-112 2061.27452-2070.73384
IR 192 110-113 2051.8152-2080.19316
IR 193 111-112 2061.27452-2070.73384
IR 193 245-246 3328.8234-3338.28272
IR 219 428 5059.87896
IR 227 211 3007.20652
IR 228 79 1758.57628
IR 228 222 3111.25904
IR 229 116 2108.57112
IR 234 175 2666.671
IR 235 175 2666.671
IR 235 226 3149.09632
IR 236 186 2770.72352
IR 237 129 2231.54228
IR 238 38 1370.74416
IR 241 233 3215.31156
IR 243 202 2922.07264
IR 244 228 3168.01496
IR 245 191-192 2818.02012-2827.47944
IR 250 414 4927.44848
===========================================================================
[ VIR DETECTOR DEFECTIVE PIXELS ]
Note that for the VIS detector, pixels located at these coordinates are to be considered both as filter boundaries and defective:
11. REFERENCES
[1] E. Ammannito, PhD dissertation, Università degli studi di Padova, Centro Interdipartimentale di Studi e Attività Spaziali (CISAS), 2008. Available on line at: http://paduaresearch.cab.unipd.it/760/1/tesi_online.pdf (in Italian)
[2] E. Ammannito, G. Filacchione, A. Coradini, F. Capaccioni, G. Piccioni, M.C. De Sanctis,M. Dami, A. Barbis, Rev. Sci. Instrum. 77, 093109 (2006)
[3] M.C. De Sanctis • A. Coradini • E. Ammannito • G. Filacchione • M.T. Capria • S. Fonte • G. Magni • A. Barbis • A. Bini • M. Dami • I. Ficai-Veltroni • G. Preti • VIR Team, 2011. The VIR Spectrometer. Space Sci Rev
DOI 10.1007/s11214-010-9668-5[4] Melchiorri, R., Piccioni, G., Mazzoni, A., 2003. Review of Scientific Instruments, vol. 74, number 8, 3796-3801.
[5] G. Filacchione, PhD dissertation, Universita degli studi di Napoli Federico II, 2006. (in Italian)
[6] G. Filacchione, E. Ammannito, A. Coradini, F. Capaccioni, G. Piccioni, M.C. De Sanctis, M. Dami, A. Barbis, Rev. Sci. Instrum. 77, 103–106 (2006)
[7] M. C. De Sanctis , J.-Ph. Combe, E. Ammannito, E. Palomba, A. Longobardo, T. B. McCord, S. Marchi, F. Capaccioni, M. T. Capria, D. W. Mittlefehldt, C. M. Pieters, J. Sunshine, F. Tosi, F. Zambon, F. Carraro, S. Fonte, A. Frigeri, G. Magni, C. A. Raymond, C. T. Russell, and D. Turrini, Detection of widespread hydrated materials on Vesta by the VIR imaging spectrometer on board the Dawn Mission, The Astrophysical Journal Letters, 758:L36 (5pp), 2012 October 20.
[8] F.G. Carrozzo A. Raponi, M. C. De Sanctis, E. Ammannito, M. Giardino, E. D'Aversa, S. Fonte, and F. Tosi, Artifacts reduction in VIR/Dawn data, Rev. Sci. Instrum., 87, Issue 12 (2016)