DATA_SET_DESCRIPTION |
Data Set Overview
=================
This volume contains the CRISM Targeted Empirical Record (TER) archive,
a collection of multiband image cubes derived from targeted (gimbaled)
observations acquired by the Compact Reconnaissance Imaging Spectrometer
for Mars (CRISM) instrument on the Mars Reconnaissance Orbiter (MRO)
spacecraft. The primary TER product (IF) consists of spectral
reflectance (I/F) data corrected for photometric, atmospheric, and
instrumental effects, with an associated text file that contains
wavelength information (WV) for each spectral band. For observations
for which both VNIR (S-detector) and IR (L-detector) hyperspectral
image cubes were acquired, the data from VNIR detectors has been
spatially registered and concatenated to the IR data to form a full
spectral range image cube. The TER product set also includes a spectral
summary parameter (SU) image cube calculated using an updated version of
the CRISM parameter library [VIVIANO-BECKETAL2014], a refined spectral
summary parameter (SR) image cube which is a noise-remediated version of
the SU product, a series of browse product (BR) spectral content
visualizations which are composed of thematically related summary
parameters, and a set of data processing information (IN) maps that
illustrate spatial/spectral residuals originating from the VNIR/IR
spectral concatenation and empirical atmospheric correction.
The constituent products can be identified by the filename/product ID
activity string prefix and the product type string:
CCCNNNNNNNN_XX_AAAAAD_TTTV
CCC = Class Type (e.g. FRT, HRL, HRS)
NNNNNNNN = Hexadecimal observation ID
XX = Hexadecimal observation segment counter
AAAAA = Activity string - composed of a 2-character prefix followed by
3-digit macro ID. The 2-character prefix describes the contents
of the file, and includes
IF - Corrected I/F
SU - Spectral summary parameters
SR - Refined spectral summary parameters
BR - Browse product
IN - Data processing information
WV - Wavelength information for the corrected I/F image
The 3-digit macro ID identifies the instrument macro that was
being run to collect the central scan part of the observation
that was processed and is represented in the product
D = sensor ID (S = VNIR; L = IR; J = Joined (VNIR + IR))
TTT = Product type string
TER - Targeted Empirical Record
V = Version number
In the TER data product set the version number (V) is the
radiometric calibration version number inherited from the source TRDRs,
and the product type string (TTT) is TER for Targeted Empirical
Record. For browse product (BR) files, the 3-digit macro ID is replaced
with a 3-character product identifier.
All TER image data are in the IR (L-detector) sensor space (not
map-projected). The image data are stored as multi-band images
(.IMG) with associated PDS labels (.LBL) and ENVI headers (.HDR). The
wavelength information is stored in text files (.TAB) also with
associated PDS labels (.LBL). The browse products (BR) are
stored in the Portable Network Graphics (.PNG) format with
alpha channel transparency, referenced by the associated PDS
labels (.LBL). TER browse products also include the scaled data values
of the .PNG file in a PDS image file (.IMG) and have an associated
ENVI header file (.HDR), allowing the user to load the PDS image file
into the ENVI software application.
Processing
==========
The Map-projected Targeted Reduced Data Record (MTRDR) data processing
pipeline produces both TERs and MTRs. It integrates a series of
standard and empirical spectral corrections, spatial transforms,
parameter calculations, and renderings in the generation of a high level
suite of analysis and visualization products. The TER/MTR data
processing flow is illustrated in Figure 2-9a in the CRISM Data Product
SIS and the TER/MTR data processing pipeline is described in detail in
the CRISM Data Product SIS Appendix P1. The CRISM TERs and MTRs are
paired data product types where an MTR is a map-projected version of a
TER. The only exception to pairing of the products is that the MTR
suite contains a DE product which is a map-projected version of the IR
(L-detector) DDR associated with the source targeted observation.
All of the TER products are in the IR (L-detector) sensor
space, so the same map-projection transformation is used to generate all
of the MTRs for a given source targeted observation. The bands in the
MTR SU, SR, BR, and IN products match the TER precursors (and the bands
in the MTR DE product match the DDR precursor). The TER IF products also
contain all the channels in the source VNIR and IR TRDRs, whereas
spectral channels with questionable radiometry ('bad bands') are
not propagated from the TER IF to the MTR IF product (thus there are
fewer spectral channels in the MTR spectral image cube as compared to
the TER precursor).
The TER/MTR data processing procedures through the generation of the
TER product set are summarized below. Detailed descriptions are provided
in Appendix P1 of the CRISM Data Product SIS.
1. Lambertian Photometric Correction (PHT)
The spectrum of each spatial pixel is divided by the cosine of the
incidence angle at that pixel.
2. Modified 'Volcano Scan' Atmospheric Correction (ATM) (IR only)
The spectrum of each spatial pixel is divided by an empirically
derived atmospheric transmission spectrum scaled to match the
depth of the 2000-nm CO2 absorption. The correction is derived
from CRISM nadir-pointed observations crossing the full range of
relief of Olympus Mons. To account for small wavelength shifts of
the IR detector over the mission, the 'best' out of a series of
temporally variable corrections is selected from a menu based
on minimizing the residual from the correction. In addition, not
all parts of the 2000-nm CO2 absorption - actually an overlapping
series of narrower absorptions - scale the same with path length.
An additional correction of the resulting artifact near 2070 nm
is superimposed to mitigate the artifact.
3. Ratio Shift Correction (RSC) (IR only)
A statistically robust 'destriping' procedure is applied to
the ATM-corrected image cube to mitigate the reintroduction of
spatial column-oriented residuals, as viewed in IR detector
space. These originate from spurious values in either ground
calibration data or flight scene or calibration data, and were
filtered out of the IF version of the source TRDR in a
statistically robust fashion. However some are reintroduced at
a low level when the empirically based ATM correction is
applied. An analogous procedure to that used in the IF version
of the source TRDR is reapplied to correct reintroduced artifacts.
4. Empirical Geometric Normalization (EGN)
Systematic brightness variations in the along-track direction of
a CRISM targeted observation result from the continuously varying
observation geometry (gimbal motion) that is used to take out
along-track motion of the field of view. The variable phase
angle and atmospheric path lengths that result from this
procedure modulate the fractions of radiance at sensor that are
contributed by the surface and atmosphere as a function of
wavelength. The effects are characterized as a function of the
continuously variable observation geometry along-track, and
normalized to the geometry at the frame of the observation which
is closest to nadir.
5. Empirical Smile Correction (ESC)
Spectral smile is an optical artifact whereby a single wavelength
does not translate uniquely to a single row of detector elements
on either the VNIR or IR detector. As a result, the wavelength of
a single detector row - which translate into a single band of a
multiband image file - drifts up and down across the field of
view. Where at-sensor radiance changes as a function of
wavelength, the sampled radiance also therefore systematically
changes. Spectral smile effects are mitigated using a fit of
along-track-averaged cross-track variation in radiance that is
constrained to have a form consistent with spectral smile
effects. Cross-track variation is normalized to the 100-colunm
wide strip near the center of the detector that corresponds to
the reference wavelength sampling vector (also called the 'SW'
CDR, in the CDR directory of the CRISM EDR archive).
6. VNIR/IR Sensor Space Transform (XFM)
The focal lengths of the VNIR and IR parts of the CRISM
instrument are slightly different, resulting in about a 1 percent
difference in pixel scale. Post-processing is required to
register the two parts of the data set; however, how they map onto
Mars' surface is known very accurately. The corrected VNIR
spectral data are remapped into the correct location in IR sensor
space based on the known surface intercepts of each spatial
pixel, using a nearest-neighbor sampling. The spectral
concatenation of the transformed corrected VNIR data and the
corrected IR data results in a fully corrected, full spectral
range data product (a TER IF image cube, accompanied by a
TER WV wavelength table).
7. Data Processing Information Generation (INF)
Residuals from the joining of the VNIR and IR data around 1000 nm
and from correction for atmospheric CO2 around 2000 nm are
parameterized in the TER IN image cube, to identify for data
users the wavelengths and spatial locations of processing artifacts.
There is also traceability back to line and sample coordinates
in the source VNIR and IR image cubes. The bands in the IN
product consist of:
'VNIR/IR Spectral Continuity Residual' - the difference in corrected
I/F between the VNIR and IR data at the wavelength of the join.
'VNIR/IR Spatial Gradient Residual' - the difference in the
sampling effects between the VNIR and IR, which can create a
spectral artifact, measured as the dot product of the normals to
a brightness gradient fitted to 3x3 spatial pixel kernels at
wavelengths above and below that of the join.
'ATM Correction Spectral Shift Artifact' - a measure of the
magnitude of a ringing-like artifact of correction for the
atmospheric CO2 bands near 2000 nm that arises when the
wavelength calibrations of the source IR data and the IR
data used to derive the atmospheric correction spectrum
are slightly misaligned, typically by a fraction of a
nanometer.
'VNIR Sample' - sample coordinate of the VNIR part of the
TER spectrum in the source VNIR TRDR.
'VNIR Line - line coordinate of the VNIR part of the
TER spectrum in the source VNIR TRDR.
'IR Sample' - sample coordinate of the IR part of the
TER spectrum in the source IR TRDR.
'IR Line' - line coordinate of the IR part of the
TER spectrum in the source IR TRDR.
'VNIR/IR Offset' - the difference in meters between the centers
of the corresponding pixels of the VNIR and IR source
products projected onto the Martian surface.
'VNIR/IR Mask' - a boolean indicator of which pixels are
populated from the VNIR and IR source products, where 0 =
populated and 1 = not populated.
8. Spectral Summary Parameter Generation (SUM)
A suite of mineral indicators and other measures of spectral
shape and reflectivity, collectively called spectral summary
parameters, is calculated from the TER IF data using the revised
and expanded spectral summary parameter library of
[VIVIANO-BECKETAL2014] and stored in the SU product. The bands in
the SU and refined summary parameter (SR) image cube are given below
along with a brief description of their significance. Users are
referred to Table 3-12 of the CRISM Data Product SIS for detailed
formulations and caveats.
R770 = 0.77-micron reflectance (higher value more dusty
or icy)
RBR = Red/blue ratio (higher value indicates more
nanophase iron oxide or sky illumination)
BD530_2 = 0.53-micron band depth (higher value has more
fine-grained crystalline hematite)
SH600_2 = 0.6-micron shoulder height (select ferric
minerals esp. hematite, goethite, or
compacted texture)
SH770 = 0.77-micron shoulder height (select ferric
minerals, less sensitive to LCP than SH600_2)
BD640_2 = 0.64-micron band depth (select ferric minerals,
esp. maghemite, but obscured by VNIR detector
artifact)
BD860_2 = 0.86-micron band depth (select crystalline
ferric minerals, esp. hematite)
BD920_2 = 0.92-micron band depth (crystalline ferric minerals
and low-Ca pyroxene, or LCP)
RPEAK1 = Reflectance peak 1 near 0.77 microns (<0.75
suggests olivine, 0.75 pyroxene, >0.8 dust)
BDI1000VIS = 1-micron integrated band depth; VNIR wavelengths
(olivine, pyroxene, or Fe-bearing glass)
BDI1000IR = 1-micron integrated band depth; IR wavelengths
(crystalline Fe2+ silicates)
R1330 = IR albedo
BD1300 = 1.3-micron absorption associated with Fe2+
substitution in plagioclase
OLINDEX3 = Broad absorption centered at 1 micron
(olivine strongly >0, also Fe-phyllosilicate)
LCPINDEX2 = Broad absorption centered at 1.81 micron
(pyroxene is strongly +; favors LCP)
HCPINDEX2 = Broad absorption centered at 2.12 microns
(pyroxene is strongly +; favors HCP)
VAR = 1.0-2.3-micron spectral variance
ISLOPE1 = Spectral slope 1 (from 1.185 to 2.530 microns;
ferric coating on dark rock)
BD1400 = 1.4-micron H2O and -OH band depth
(hydrated or hydroxylated minerals)
BD1435 = 1.435-micron CO2 ice band depth
BD1500_2 = 1.5-micron H2O ice band depth
ICER1_2 = CO2 and H2O ice band depth ratio
BD1750_2 = 1.75-micron H2O band depth (gypsum or alunite)
BD1900_2 = 1.9-micron H2O band depth (hydrated minerals
except monohydrated sulfates)
BD1900r2 = 1.9-micron H2O band depth (hydrated minerals
except monohydrated sulfates)
BDI2000 = 2-micron integrated band depth (pyroxene)
BD2100_2 = 2.1-micron shifted H2O band depth
(monohydrated sulfates)
BD2165 = 2.165-micron Al-OH band depth
(pyrophyllite, kaolinite-group minerals)
BD2190 = 2.190-micron Al-OH band depth
(beidellite, allophane, imogolite)
MIN2200 = 2.16-micron Si-OH band depth and 2.21-micron
H-bound Si-OH band depth (doublet; kaolinite)
BD2210_2 = 2.21-micron Al-OH band depth (Al-OH minerals)
D2200 = 2.2-micron dropoff (Al-OH minerals)
BD2230 = 2.23-micron band depth
(hydroxylated ferric sulfate)
BD2250 = 2.25-micron broad Al-OH and Si-OH band depth
(opal, Al-OH minerals)
MIN2250 = 2.21-micron Si-OH band depth and 2.26-micron
H-bound Si-OH band depth (opal)
BD2265 = 2.265-micron band depth (jarosite, gibbsite,
acid-leached nontronite)
BD2290 = 2.29-micron Mg,Fe-OH band depth / 2.292-micron
CO2 ice band depth (Mg-OH and Fe-OH minerals,
Mg carbonate, and CO2 ice)
D2300 = 2.3-micron dropoff (hydroxylated Fe,Mg
silicates strongly >0)
BD2355 = 2.35-micron band depth (chlorite, prehnite,
pumpellyite, carbonate, serpentine)
SINDEX2 = Inverse lever rule to detect convexity at 2.29
microns due to 2.1- and 2.4-micron absorptions
(hydrated sulfates strongly >0)
ICER2_2 = 2.7-micron CO2 ice band
MIN2295_2480 = Mg Carbonate overtone band depth and metal-OH
band
MIN2345_2537 = Ca/Fe Carbonate overtone band depth and
metal-OH band
BD2500_2 = Mg Carbonate overtone band depth
BD3000 = 3-micron H2O band depth (adsorbed and bound
H2O and ice)
BD3100 = 3.1-micron H2O ice band depth
BD3200 = 3.2-micron CO2 ice band depth
BD3400_2 = 3.4-micron carbonate band depth
CINDEX2 = Inverse lever rule to detect convexity at 3.6
micron due to 3.4- and 3.8-micron
carbonate absorptions
R440 = 0.44-micron reflectance
R530 = 0.53-micron reflectance
R600 = 0.60-micron reflectance
IRR1 = IR ratio 1 (R880/R997; aphelion ice clouds >1,
seasonal ice clouds and dust <1))
R1080 = 1.08-micron reflectance
R1506 = 1.51-micron reflectance
R2529 = 2.53-micron reflectance
BD2600 = 2.6-micron H2O band depth
IRR2 = IR ratio 2 (R2530/R2210; aphelion ice clouds
vs. seasonal ice clouds or dust)
IRR3 = IR ratio 3 (R3500/R3390; aphelion ice clouds
vs. seasonal ice clouds or dust)
R3920 = 3.92-micron reflectance, evaluated from the
5 'good bands' closest in wavelength to
3920 nm
9. Refined Spectral Summary Parameters (SRE)
The majority of the parameter bands are filtered using a variant of
the filtering algorithm used on IR TRDR I/F image cubes, to
mitigate spurious noise structure that remains and is accentuated
in the summary products. The noise remains because the filtering
applied to the TRDRs uses a conservative threshold for calling an
outlying pixel 'noise' and interpolating through it, to avoid
altering the actual information content of the data. Low-magnitude
noise that 'leaks though' TRDR processing appears larger in some
of the summary products prior to filtering because their dynamic
range is typically small compared to the I/F data.
10. Browse Product Generation (BRS)
Browse products are greyscale or RGB composites of 1 or 3
thematically related summary products remapped to 8 bits that allow
for a quick assessment of the information content of the source
spectral data TER IF image cubes).
The TER browse product set consists of three data files and a
detached PDS label file. The label file contains the metadata and
pointers to the three data files. The three data files are 1) an IMG
file containing the browse product image as a three-band PDS IMAGE
object; 2) a PNG file containing the browse product image in the
Portable Network Graphics file format (three bands and an alpha
transparency channel); and 3) a HDR file associated with the IMG file
in the ENVI header format. This allows users of the ENVI image
processing software to readily read in the image data. The
SOURCE_PRODUCT_ID keyword in the PDS label links the browse products
to the source TER.
The following formulations are used.
Abbreviation = TRU
Name = True color
R Component = R600
G Component = R530
B Component = R440
Additional description: An enhanced natural color representation
of the scene composed of spectral channels across the visible
spectrum. Contrast greater than the human eye would see.
Abbreviation = VNA
Name = VNIR albedo
R Component = R770
G Component = R770
B Component = R770
Additional description: Reflectance at 770 nm as a proxy for VNIR
albedo and may be used to correlate spectral variations with
morphology.
Abbreviation = FEM
Name = Fe Minerals
R Component = BD530_2
G Component = SH600_2
B Component = BDI1000VIS
Additional description: Mafic minerals appear blue, nanophase
ferric oxides red, and dust-coated mafic rocks or lithified
dust yellow/green.
Abbreviation = FM2
Name = Fe minerals, v2
R Component = BD530_2
G Component = BD920_2
B Component = BDI1000VIS
Additional description: Mafic minerals appear blue, crystalline
hematite green or yellow, and nanophase ferric oxides red.
Abbreviation = TAN
Name = Tandem
R Component = R2529
G Component = IRA
B Component = R770
Additional description: Enhanced visible to infrared false color
representation of the scene, incorporating spectral data from
both (VNIR and IR) detectors.
Abbreviation = IRA
Name = IR albedo
R Component = R1300
G Component = R1300
B Component = R1300
Additional description: Reflectnace at 1330 nm as a proxy for IR
albedo and may be used to correlate spectral variations with
morphology.
Abbreviation = FAL
Name = False color
R Component = R2529
G Component = R1506
B Component = R1080
Additional description: Red to orange coloration is
typically characteristic of olivine-rich material, blue/green
colors are often indicative of clay mineralogy, green colors
may indicate carbonate, and gray/brown colors often indicate
basaltic material.
Abbreviation = MAF
Name = Mafic mineralogy
R Component = OLINDEX3
G Component = LCPINDEX2
B Component = HCPINDEX2
Additional description: Olivine and Fe-phyllosilicate share
a 1000-1700 nm bowl-shaped absorption and will appear red in
the MAF browse product. Low- and high-Ca pyroxene display
an additional ~2000-nm absorption and appear green/cyan and
blue/magenta respectively.
Abbreviation = HYD
Name = Hydrated mineralogy
R Component = SINDEX2
G Component = BD2100_2
B Component = BD1900_2
Additional description: Polyhydrated sulfates have strong
1900 nm and 2400 nm absorption bands, and thus appear magenta
in the HYD browse product. Monohydrated sulfates have a
strong 2100 nm absorption and a weak 2400 nm absorption band,
and thus appear yellow/green in the HYD browse product.
Blue colors are indicative of other hydrated minerals (such as
clays, glass, carbonate, or zeolite).
Abbreviation = PHY
Name = Phyllosilicates
R Component = D2300
G Component = D2200
B Component = BD1900r2
Additional description: Fe/Mg-OH bearing minerals
(e.g., Fe/Mg-phyllosilicate) will appear red, or magenta
(when hydrated). Al/Si-OH bearing minerals (e.g.,
Al-phyllosilicates or hydrated silica) will appear green,
or cyan (when hydrated). Blue colors are indicative of other
hydrated minerals (such as sulfates, glass, carbonate, or
water ice).
Abbreviation = PFM
Name = Phyllosilicates with Fe and Mg
R Component = BD2355
G Component = D2300
B Component = BD2290
Additional description: Red/yellow colors indicate the
presence of prehnite, chlorite, epidote, or Ca/Fe carbonate,
while cyan colors indicate the presence of Fe/Mg-smectites
or Mg-carbonate.
Abbreviation = PAL
Name = Phyllosilicates with Al
R Component = BD2210_2
G Component = BD2190
B Component = BD2165
Additional description: Red/yellow colors indicate the
presence of Al-smectites or hydrated silica, cyan colors may
indicate the alunite, and light/white colors indicate the
presence of kaolinite group minerals.
Abbreviation = HYS
Name = Hydrated silica
R Component = MIN2250
G Component = BD2250
B Component = BD1900r2
Additional information: Light red/yellow colors indicate
the presence of hydrated silica, whereas cyan colors indicate
Al-OH minerals. Additionally, jarosite will appear yellow.
Blue colors are indicative of other hydrated minerals (such
as sulfates, clays, glass, carbonate, or water ice).
Abbreviation = ICE
Name = Ices
R Component = BD1900_2
G Component = BD1500_2
B Component = BD1435
Additional information: CO2 frost or ice displays a sharp
1435-nm absorption and thus appears blue in the ICE browse
product. Water ice or frost has a strong 1500 nm absorption and
thus appears green in the ICE browse product. Red colors are
indicative of hydrated minerals (such as sulfates, clays,
glass, carbonate, or water ice).
Abbreviation = IC2
Name = Ices, v2
R Component = R3920
G Component = BD1500_2
B Component = BD1435
Additional information: CO2 frost or ice displays a sharp
1435-nm absorption and thus appears blue in the IC2 browse
product. Water ice or frost has a strong 1500 nm absorption and
thus appears green in the IC2 browse product. The 3920-nm
spectral channel is a discriminator for icy vs. ice-free material
with ices having a lower solar reflected and thermal emission
radiance at this wavelength, so ice-free material appears red.
Abbreviation = CHL
Name = Chloride
R Component = ISLOPE
G Component = BD3000
B Component = IRR2
Additional information: Martian chloride deposits have a
relatively positive near-infrared spectral slope and are
comparatively desiccated, so chlorides appear blue in the
CHL browse product. Yellow/green colors are indicative of
hydrated minerals, especially phyllosilicates.
Abbreviation = CAR
Name = Carbonates
R Component = D2300
G Component = BD2500H2
B Component = BD1900_2
Additional information: Blueish- or yellowish-white colors
indicate Mg-carbonate, while red/magenta colors indicate
Fe/Mg-phyllosilicate. Blue colors are indicative of other
hydrated minerals (such as sulfates, clays, glass, or
carbonate).
Abbreviation = CR2
Name = Carbonates, v2
R Component = MIN2295_2480
G Component = MIN2345_2537
B Component = CINDEX2
Additional information: Red/magenta colors indicate
Mg-carbonates, while green/cyan colors indicate
Fe/Ca-carbonates.
Extras
======
The Targeted Empirical Record (TER) EXTRAS directory contains a series
of data processing visualizations for each CRISM hyperspectral targeted
observation that has been processed through the Map-projected Targeted
Reduced Data Record (MTRDR) pipeline. These visualizations collectively
depict the geometric structure of the source TRR3 I/F spectral data,
illustrate the spatial and spectral impact of the TER/MTRDR data
processing procedures, and provide snapshots of the underlying modeling
behavior of the empirical data processing. All TER/MTRDR EXTRAS
visualizations are Portable Network Graphics (PNG) format files.
Detailed descriptions of the TER/MTRDR EXTRAS products are provided in
the CRISM Data Product SIS Appendix P2.
Limitations
===========
None.
|
CONFIDENCE_LEVEL_NOTE |
By design many of the sources of
uncertainty in interpretation of the data relevant to calibrated
data or TRDRs are reduced or eliminated in TERs. Remaining issues
of most concern to data users follow.
(1) Variable spectral resolution
In order to distinguish spectrally similar minerals that
have different geological implications for their environments
of formation, adequate spectral resolution is necessary. This
requires sufficiently high density spectral sampling, as well
as a sufficiently narrow full width half maximum (FWHM) of the
instrument response in the spectral direction. This 'slit
function,' the effective bandpass for a single detector
element, represents the convolution of spectral sampling and
the point-spread function in the spectral direction. CRISM's
benchmark is distinguishing the minerals montmorillonite and
kaolinite, which form in hydrothermal environments under
different temperature regimes [SWAYZEETAL2003]. The requirements
for this are (a) <20 nm FWHM and (b) sampling of the spectrum
at this or smaller increments. CRISM's spectral sampling
requirement is <10 nm/channel to provide oversampling, and the
actual performance is better at 6.55 nm/channel. FWHM is 8 nm
in the VNIR across the FOV. In the IR it increases from 10 nm
at short wavelengths to 15 nm at the longest wavelengths at the
center of the FOV, and broadens by about 2 nm at 0.8 degrees
from the center of the field of view. Outside +/-0.9 degrees
from the center of the field of view the telescope is slightly
vignetted, so further degradation is expected at extreme field
angles. Although the spectral sampling and resolution meet
requirements, their variation across the field-of-view must be
accounted for when comparing with rock and mineral analog
spectra. The FWHM of the slit function is given for the reference
columns of the source TRDR as part of the WV table accompanying
the TER.
(2) Long-wavelength calibration uncertainty
The responsivity correction at IR wavelengths 3000-3920 is suspected
to contain low wavelength frequency errors, perhaps leading to a
broad 'bump' centered near 3400 nm. This is currently under
investigation and may be corrected in a future version of the IR
radiometric calibration.
(3) Residuals from correction of the 2000-nm CO2 gas absorption.
As described above, parts of many scenes contain a ringing-like
artifact of using a correction for atmospheric gases derived from
data where the IR wavelength calibration has drifted slightly.
An indicator of parts of an image subject to this effect is
included in the informational IN image cube.
(4) Artifacts near 1 micron
For reasons that are not well understood, the quality of the
calibration of VNIR and IR data on either side of the 'join' near
1 micron drifts in time. The IN informational image cube contains
indicators of where in an image this is most severe. The calibration
of data away from the join is only suspect where elsewhere indicated.
(5) 'Bad bands'
The TER data product contains all constituent bands from the source
VNIR and IR data products. Wavelength near the limits of optical
zones in the instrument typically have less reliable calibration and
should be routinely ignored if possible. Other wavelengths exhibit
occasional, non-persistent calibration artifacts. These include:
Wavelengths less than 442 nm (due to artifacts from the scattered
light correction in high contrast scenes).
Wavelengths between 631 and 710 nm (due to optical artifacts at the
boundary between zones of the VNIR detector).
Wavelengths betweeen 970 nm and 1047 nm have calibration that varies
between observations; the reason is uncertain but may be related to
uncorrected effects of beamsplitter temperature).
In some observations, there is a column-dependent artifact in the
form of a broad dip near 1220 nm, whose origin is unknown. The
artifact disappears when a spectrum containing it is ratioed to
spectrally bland material within the same observation, in or near
the source IR detector column.
In some observations, there is a spike or trough at the boundary
between the short and intermediate wavelength segments of the IR
detector, near 1660 nm. The artifact disappears when a spectrum
containing it is ratioed to spectrally bland material within the
same observation, in or near the source IR detector column.
Wavelengths 2660-2800 nm (the reason is uncertain but may be due
to problems with correction of water vapor in measurements of the
ground calibration sources).
The shape of the spectrum at 3100-3800 nm is suspect and there
may be a broad, low 'bump'.
(6) Scene-dependent opacity of dust and ice aerosols
The processing that normalizes atmospheric effects does not attempt
to remove scattering effects of dust and ice aerosols, only to
normalize them to the effects at the nearest-to-nadir geometry in
the scene in question. Scenes with high opacities of dust or ice, for
example observed during global dust events, are by design excluded
from the TER archive. Thresholds used were dust opacity (tau) > 1.39
and ice opacity (tau) > 0.28. However overlapping scenes within the
archive may have different dust or ice loads below these limits, so
overlapping spectra measured at different times may have distinct
values. The effects will be greatest at shorter wavelengths,
and in absorptions related to iron minerals. In addition some scenes
are observed through a thin water ice haze. These will have
characteristic weak ice absorption near 1500 and 2000 nm.
(7) Summary product and browse product cautions
Summary products and browse products are intended to provide rapid
overviews of the content of CRISM hyperspectral data, and to convey
spatial variations in mineral spectral signatures in a compact
fashion. However they are not conclusive indicators of the presence
of particular minerals; false positives and false negatives are
not uncommon. Users are referred to Table 3-12 of the CRISM Data
Product SIS for detailed caveats regarding false positives, and to
[VIVIANO-BECKETAL2014] for an extended discussion of the topic.
|