Surface Stereo Imager (SSI) data sets

Data Set Overview

There are multiple methods of performing radiometric correction, distinguished by the RADIOMETRIC_CORRECTION_TYPE keyword. The most common are TAMCAL, RACCAL, MIPLRAD, MIPLRAD2, and MIPLRAD3. This data set has been processed with SSICAL.

1. SSICAL Method (SSI Team)

This refers to radiometric correction of SSI instrument data only, performed by the SSI instrument team (Texas A&M University and University of Arizona) using their suite of software tools. It is the most precise correction method applicable to SSI data.

There are 2 general types of SSI Radiometrically-corrected RDR products that are generated by the SSI instrument team: Radiance-calibrated and Radiance-factor calibrated. Additional details on the radiometric processing and calibration of SSI images can be found in the SSI Calibration Report.

1.1. Radiance-calibrated RDRs ('RAD')

The non-linearized RDRs are generated from EDRs. They have all of the major instrumental/environmental calibrations applied, such as bias removal, dark current removal, electronic shutter smear effect removal, flat field correction, and bad pixel repair. Then they have been scaled to absolute radiance units using pre-flight radiometric calibration coefficients. The units on these files are (W/m^2/nm/sr). In addition, floating point versions of this RDR may also be generated.

1.2. Radiance factor-calibrated RDRs ('IOF')

The non-linearized RDRs are generated from EDRs or 'RAD' RDRs. They have all the major instrumental/environmental calibrations applied and have been scaled to absolute radiance units as described above, and then have been divided by the absolute radiance of the Sun at the top of the Martian atmosphere within the appropriate SSI bandpass, to generate radiance factor, or 'I over F' values, where I is the radiance from the Martian scene and pi * F is the radiance from the Sun at the top of the Martian atmosphere (or on the surface, as determined by reflectance calibration targets. Since the solar radiance in the same units as the Mars scene radiance was divided out, these files are unitless but typically have values in the range of 0.0 to 1.0 (for example, average bright Mars soils exhibit I/F ~ 0.35 at 750 nm and I/F ~ 0.05 at 410 nm). These files are further corrected with reference to the SSI Reflectance Calibration Targets (RCTs) to make the result I/F where F has been modified to the flux incident on the surface at the bottom of the atmosphere. To get R*, divide the IOF data by the cosine of the solar zenith angle. As with the 'RAD' RDR type, there exists a linearized version of the IOF type of radiometrically corrected RDR, called 'IOL'. A floating point version of this RDR may also be generated.

2. RACCAL Method (RAC Team)

This refers to radiometric correction of RAC instrument data only, performed by the RAC instrument team (MPS) using their suite of RACCAL software tools. It is the most precise correction method applicable to RAC data. Note that radiometric correction of MECA-OM instrument data will be performed using the same tools employed for the RACCAL method.

The RAC/OM calibration steps performed by the RACSoft package are described below:

1. The bad pixel removal state replaces a number of pixels marked bad because of dust grains on the CCD or hot electron production. The bad pixels are replaced by an interpolated value based on the surrounding pixels.

2. The bias subtraction state subtracts the ADC digital offset from the image.

3. The RAC and the OM uses an electronic shutter where the image data is fast clocked to a covered area on the CCD at the end of the exposure. During the fast clocking each row experiences addition light from other parts of the scene. The electronic shutter correction subtracts from row N the summed DN signal of row 0 to N-1 scaled by the time it takes to clock a row one step on the CCD.

4. The dark current correction subtracts an estimated mean value of dark current based on the temperature of the CCD. This simple scheme (as compared to the SSI) is used because the RAC and OM has a very low dark current production under Mars conditions.

5. The flatfield correction divider the image by the relevant flatfield for the given focus motor step.

6. The OM calibration is finished after the flat field correction since good absolute calibration data is not available for the OM.

7. The final step of the RAC calibration is to divide the image by the absolute calibration constant for the given focus motor step. The calibration constant is given by the ground absolute calibration at focus motor step 306 (near infinite focus) and a correction factor derived for the change in instantaneous field of view between focus step 306 and the active focus step.

3. MIPLRAD, MIPLRAD2, MIPLRAD3 Methods

These refer to radiometric correction of any camera instrument data systematically performed by MIPL (OPGS at JPL) to meet tactical time constraints imposed by Arm Planners. The resulting rad-corrected RDRs are integrated into terrain mesh products used for RA trench digging. For SSI and RAC instrument data, these methods are less precise than the SSICAL and RACCAL methods previously discussed. The MIPL radiometrically-corrected RDR filenames carry the product type designators RAD (non-linearized) or RAL (linearized).

MIPLRAD, MIPLRAD2, and MIPLRAD3 are first-order corrections only and should be considered approximate. All three apply the following corrections: dark current, temperature-compensated responsivity, exposure time, binning correction, and flat field. The result is calibrated to physical units for PHX of W/m^2/nm/sr. The actual algorithm and equations used for the three MIPLRAD's are in the data product SIS [ALEXANDERETAL2008]. In all cases, ALL_CAPITALS serve to denote keyword names in the PDS label.

The only difference between the three MIPLRAD methods is in the dark current calculation that is used. MIPLRAD uses a dark current calculation developed by Adam Shaw at the University of Arizona. MIPLRAD2 (the default) uses a calculation developed by Mark Lemmon at Texas A&M University. MIPLRAD3 uses the Lemmon calculation with a simplification for efficiency (described in the data product SIS).

Dark current applies only to SSI. RAC dark current is assumed to be 0 in all three methods.

Processing

Phoenix SSI RDRs are considered Level 3 (Calibrated Data equivalent to NASA Level 1-A), Level 4 (Resampled Data equivalent to NASA Level 1-B), or Level 5 (Derived Data equivalent to NASA Level 1-C, 2 or 3). The RDRs are to be reconstructed from Level 2 edited data, and are to be assembled into complete images that may include radiometric and/or geometric correction.

Phoenix SSI instrument EDRs and RDRs will be generated by JPL's Multimission Instrument Processing Laboratory (MIPL) as part of the OPGS subsystem of the Phoenix GDS. RDRs will also be generated by the SSI science instrument team at the SOC facility at the University of Arizona, as well as at its home institution, Texas A&M.

RDR data products will be generated by, but not limited to, MIPL using the Mars Suite of VICAR image processing software at JPL, and the SSI science instrument team using TAMCAL and RACCAL software at the SOC facility at the University of Arizona and at the team's home institution at Texas A&M University. The RDRs produced will be 'processed' data. The input will be one or more Camera EDR or RDR data products and the output will be formatted according to the data product SIS [ALEXANDERETAL2008]. Additional meta-data may be added by the software to the PDS label.

Data

RDR products generated by MIPL will have a VICAR label wrapped by a PDS label, and their structure can include the optional EOL label after the binary data. RDR products not generated by MIPL may contain only a PDS label. Or, RDR products conforming to a standard other than PDS, such as JPEG compressed or certain Terrain products, are acceptable without a PDS header during mission operations, but may not be archivable.

The RDR data product is comprised of radiometrically decalibrated and/or camera model corrected and/or geometrically altered versions of the raw camera data, in both single and multi-frame (mosaic) form. Most RDR data products will have PDS labels, or if generated by MIPL (OPGS), dual PDS/VICAR labels. Non-labeled RDRs include JPEG compressed products and the Terrain products.

Software

The MIPL Mars Program Suite was used to generate these RDRs, as well as the TAMCAL and RACCAL software suites.

Media/Format

The data set will initially be delivered and kept online. Upon Mission completion, the SSI Operations RDRs will be delivered to PDS on DVD.

Data Set Overview

This data set contains raw Surface Stereo Imager (SSI) operational data. If 12 to 8 bit scaling was commanded, these images HAVE NOT been transformed back to 12 bits. These images are only used to assess the morphology, topography, and geologic context of the lander site and should not be used for quantitative scientific purposes.

More information is found in ALEXANDERETAL2008, LEMMONETAL2007, and LEMMONETAL2008.

Processing

This documentation uses the Committee on Data Management and Computation (CODMAC) data level numbering system. The Phoenix Camera Payload EDRs referred to in this document are considered Level 2 or Edited Data (equivalent to NASA Level 0). The EDRs are to be reconstructed from Level 1 or Raw Data, which are the telemetry packets within the project specific Standard Formatted Data Unit (SFDU) record. They are to be assembled into complete images, but will not be radiometrically or geometrically corrected.

EDR data products will be generated by MIPL using the telemetry processing software mertelemproc at JPL. The EDRs produced will be raw uncalibrated data reconstructed from telemetry packet SFDUs and formatted according to this SIS. Meta-data acquired from the telemetry data headers and a meta-data database will be used to populate the PDS label. Missing packets will be identified and reported for retransmission to the ground as partial datasets. Prior to retransmission, the missing EDR data will be filled with zeros. The EDR data will be reprocessed only after all partial datasets are retransmitted and received on the ground. In these cases, the original EDR version will be overwritten.

Data

The data packaged in the camera data files will be decoded, decompressed camera image data in single frame form as an Experiment Data Record (EDR). The Full Frame form of a standard image data file has the maximum dimensions of 1024 lines by 1024 samples.

1. Full Frame EDR: Full Frame EDRs are stored as 16-bit signed integers. If 12-to-8 bit scaling is performed, then pixels are stored in 16-bit format and only the last 8 bits of the 16-bit integer are used.

2. Thumbnail EDR: Thumbnail EDRs are stored as 16-bit signed integers or 8-bit unsigned integers. If 12-to-8 bit scaling is performed, then pixels are stored in 16-bit format and only the last 8 bits of the 16- bit integer are used. The Thumbnail EDR is a sized down version of the original acquired image (i.e., camera returned pixel data), and size of the binary EDR image data is variable. However, the original acquired image is not always downlinked. The main purpose of a Thumbnail EDR is to provide an image summary using a very low data volume compared to the original image.

3. Sub-frame EDR: Sub-frame EDRs are a subset of rows and columns of the 1024 x 1024 full frame image. Sub-frame EDRs are stored as 16-bit signed integers. If 12-to-8 bit scaling is performed, then pixels are stored in 16-bit format and only the last 8 bits of the 16-bit integer are used.

4. Downsampled EDR: A downsampled EDR is a smaller version of the 1024 x 1024 full frame or subframed image using the following methods: 1) nearest neighbor pixel averaging, 2) pixel averaging with outlier rejection or 3) computing the median pixel value. Downsampled EDRs are stored as 16-bit signed integers. If 12-to-8 bit scaling is performed, then pixels are stored in 16-bit format and only the last 8 bits of the 16-bit integer are used.

5. Reference Pixels: The onboard CCD array has 16 pre-Reference dark pixels (12-bits) located at the beginning and 15 post-Reference dark pixels (12-bits) located at the end of each row. The values of these pixels indicate the bias level of the camera at the time of each observation. The Reference Pixel images were losslessly compressed for downlink. For complex design reasons, the last post-Reference pixel is a copy of the next-to-last post-Reference pixel. Following the last post-Reference dark pixel, at the very end of each row, is the camera hardware serial number (left-shifted by 4 bits if 12-bit data).

Software

Phoenix Camera Payload downlink processing software is focused on rapid reduction, calibration, and visualization of images in order to make discoveries, to accurately and expeditiously characterize the geologic environment around the lander, and to provide timely input for operational decisions concerning target selection. Key software tools have been developed at JPL by the MIPL, SSV, and APSS groups, at NASA Ames, and at the USGS/Flagstaff.

PDS-labeled images and tables can be viewed with the program NASAView, developed by the PDS and available for a variety of computer platforms from the PDS web site http://pdsproto.jpl.nasa.gov/Distribution/license.html. There is no charge for NASAView.

Media/ Format

The data set will initially be delivered and kept online. Upon Mission completion, the SSI EDRs will be delivered to PDS on DVD.

Data Set Overview

There are multiple methods of performing radiometric correction, distinguished by the RADIOMETRIC_CORRECTION_TYPE keyword. The most common are TAMCAL, RACCAL, MIPLRAD, MIPLRAD2, and MIPLRAD3.

1. TAMCAL Method (SSI Team)

This refers to radiometric correction of SSI instrument data only, performed by the SSI instrument team (Texas A&M University and University of Arizona) using their suite of software tools. It is the most precise correction method applicable to SSI data.

There are 2 general types of SSI Radiometrically-corrected RDR products that are generated by the SSI instrument team: Radiance-calibrated and Radiance-factor calibrated. Additional details on the radiometric processing and calibration of SSI images can be found in the SSI Calibration Report.

1.1 Radiance-calibrated RDRs ('RAD', 'RAL')

The non-linearized RDRs are generated from EDRs. They have all of the major instrumental/environmental calibrations applied, such as bias removal, dark current removal, electronic shutter smear effect removal, flat field correction, and bad pixel repair. Then they have been scaled to absolute radiance units using pre-flight radiometric calibration coefficients. The units on these files are (W/m^2/nm/sr). An analogous RDR file type exists for the linearized (geometrically-corrected) SSI RDR as well, and it is labeled with the 'RAL' product type identifier to correspond with the 'RAD' type. In addition, floating point versions of this RDR may also be generated.

1.2. Radiance factor-calibrated RDRs ('IOF', 'IOL')

The non-linearized RDRs are generated from EDRs or 'RAD' RDRs. They have all the major instrumental/environmental calibrations applied and have been scaled to absolute radiance units as described above, and then have been divided by the absolute radiance of the Sun at the top of the Martian atmosphere within the appropriate SSI bandpass, to generate radiance factor, or 'I over F' values, where I is the radiance from the Martian scene and pi * F is the radiance from the Sun at the top of the Martian atmosphere (or on the surface, as determined by reflectance calibration targets. Since the solar radiance in the same units as the Mars scene radiance was divided out, these files are unitless but typically have values in the range of 0.0 to 1.0 (for example, average bright Mars soils exhibit I/F ~ 0.35 at 750 nm and I/F ~ 0.05 at 410 nm).

As with the 'RAD' RDR type, there exists a linearized version of the IOF type of radiometrically corrected RDR, called 'IOL'. A floating point version of this RDR may also be generated.

2. RACCAL Method (RAC Team)

This refers to radiometric correction of RAC instrument data only, performed by the RAC instrument team (MPS) using their suite of RACCAL software tools. It is the most precise correction method applicable to RAC data. Note that radiometric correction of MECA-OM instrument data will be performed using the same tools employed for the RACCAL method.

The RAC/OM calibration steps performed by the RACSoft package are described below:

1. The bad pixel removal state replaces a number of pixels marked bad because of dust grains on the CCD or hot electron production. The bad pixels are replaced by an interpolated value based on the surrounding pixels.

2. The bias subtraction state subtracts the ADC digital offset from the image.

3. The RAC and the OM uses an electronic shutter where the image data is fast clocked to a covered area on the CCD at the end of the exposure. During the fast clocking each row experiences addition light from other parts of the scene. The electronic shutter correction subtracts from row N the summed DN signal of row 0 to N-1 scaled by the time it takes to clock a row one step on the CCD.

4. The dark current correction subtracts an estimated mean value of dark current based on the temperature of the CCD. This simple scheme (as compared to the SSI) is used because the RAC and OM has a very low dark current production under Mars conditions.

5. The flatfield correction divider the image by the relevant flatfield for the given focus motor step.

6. The OM calibration is finished after the flat field correction since good absolute calibration data is not available for the OM.

7. The final step of the RAC calibration is to divide the image by the absolute calibration constant for the given focus motor step. The calibration constant is given by the ground absolute calibration at focus motor step 306 (near infinite focus) and a correction factor derived for the change in instantaneous field of view between focus step 306 and the active focus step.

3. MIPLRAD, MIPLRAD2, MIPLRAD3 Methods

These refer to radiometric correction of any camera instrument data systematically performed by MIPL (OPGS at JPL) to meet tactical time constraints imposed by Arm Planners. The resulting rad-corrected RDRs are integrated into terrain mesh products used for RA trench digging. For SSI and RAC instrument data, these methods are less precise than the TAMCAL and RACCAL methods previously discussed. The MIPL radiometrically-corrected RDR filenames carry the product type designators RAD (non-linearized) or RAL (linearized).

MIPLRAD, MIPLRAD2, and MIPLRAD3 are first-order corrections only and should be considered approximate. All three apply the following corrections: dark current, temperature-compensated responsivity, exposure time, binning correction, and flat field. The result is calibrated to physical units for PHX of W/m^2/nm/sr. The actual algorithm and equations used for the three MIPLRAD's are in the data product SIS [ALEXANDERETAL2008]. In all cases, ALL_CAPITALS serve to denote keyword names in the PDS label.

The only difference between the three MIPLRAD methods is in the dark current calculation that is used. MIPLRAD uses a dark current calculation developed by Adam Shaw at the University of Arizona. MIPLRAD2 (the default) uses a calculation developed by Mark Lemmon at Texas A&M University. MIPLRAD3 uses the Lemmon calculation with a simplification for efficiency (described in the data product SIS).

Dark current applies only to SSI. RAC dark current is assumed to be 0 in all three methods.

More information is found in ALEXANDERETAL2008, LEMMONETAL2007, and LEMMONETAL2008.

Processing

Phoenix SSI RDRs are considered Level 3 (Calibrated Data equivalent to NASA Level 1-A), Level 4 (Resampled Data equivalent to NASA Level 1-B), or Level 5 (Derived Data equivalent to NASA Level 1-C, 2 or 3). The RDRs are to be reconstructed from Level 2 edited data, and are to be assembled into complete images that may include radiometric and/or geometric correction.

Phoenix SSI instrument EDRs and RDRs will be generated by JPL's Multimission Instrument Processing Laboratory (MIPL) as part of the OPGS subsystem of the Phoenix GDS. RDRs will also be generated by the SSI science instrument team at the SOC facility at the University of Arizona, as well as at its home institution, Texas A&M.

RDR data products will be generated by, but not limited to, MIPL using the Mars Suite of VICAR image processing software at JPL, and the SSI science instrument team using TAMCAL and RACCAL software at the SOC facility at the University of Arizona and at the team's home institution at Texas A&M University. The RDRs produced will be 'processed' data. The input will be one or more Camera EDR or RDR data products and the output will be formatted according to the data product SIS [ALEXANDERETAL2008]. Additional meta-data may be added by the software to the PDS label.

Data

RDR products generated by MIPL will have a VICAR label wrapped by a PDS label, and their structure can include the optional EOL label after the binary data. RDR products not generated by MIPL may contain only a PDS label. Or, RDR products conforming to a standard other than PDS, such as JPEG compressed or certain Terrain products, are acceptable without a PDS header during mission operations, but may not be archivable.

The RDR data product is comprised of radiometrically decalibrated and/or camera model corrected and/or geometrically altered versions of the raw camera data, in both single and multi-frame (mosaic) form. Most RDR data products will have PDS labels, or if generated by MIPL (OPGS), dual PDS/VICAR labels. Non-labeled RDRs include JPEG compressed products and the Terrain products.

Software

The TAMCAL and RACCAL software was used to generate these RDRs.

Media/ Format

The data set will initially be delivered and kept online. Upon Mission completion, the SSI Operations RDRs will be delivered to PDS on DVD.

Data Set Overview

A Mars atmospheric opacity data product consists of two files, an ASCII formatted detached PDS label file and an ASCII formatted data file. The data file contains values of the Mars atmospheric opacity or optical depth derived from PHX SSI images of the Sun acquired with the two solar filters. The effective wavelengths of these filters are 451, 671, 887 and 991 n. Each data file contains an ASCII table of the derived atmospheric opacity for the given SSI solar filter. Each table contains columns with the source SSI image identifier, the time of image acquisition, the Mars season (Ls), Sun-Mars distance, airmass, observed solar flux, and opacity.

Processing

PHX SSI Atmospheric Opacity RDRs are considered Level 5 or Derived Data (equivalent to NASA Level 2). The Mars atmospheric opacity data products are generated from analysis of PHX SSI images.

The Mars atmospheric opacity data products are produced by the Surface Science Imager team using processing procedures and software developed by Mark Lemmon, Texas A&M University.

The data product is generated after each sol in which opacity data is acquired. The generation is done in four steps. First, the input parameters are set up. A list of SSI data products to be processed is read and associated values are determined for Ls, the distance of Mars from the Sun, the sol that the data were acquired, and the actual elevation angle of the Sun. It is likely that standard tools such as the NAIF toolkit will be used for these computations. For each SSI data product, the airmass is computed by integration through a spherically symmetric atmosphere with a scale height equivalent to the gas scale height of the Martian atmosphere.

Second, the solar flux is extracted from each calibrated SSI data product. To do this, the background is determined within an annulus at a fixed radius from the center of the Sun in the image. That background is subtracted, as it would lead to a significant departure from Beers' Law at high airmasses. After background subtraction, the solar flux is integrated over the image. The presence of a few missing pixels (e.g., a Phobos transit or a missing packet that only partly overlaps the Sun) can be accommodated by the integration algorithm. The presence of a large number of missing pixels or any saturated pixels will result in the rejection of an image (returning a flux and opacity of -1.000).

Third, a relative calibration is derived. Data from the afternoon of all sols during which more than 1 image was acquired are considered, together with instrumental uncertainties. The published calibration is considered as a single datum with associated uncertainty. The instrument response is varied, and a single best-fit opacity is derived for each afternoon using Beers' Law (I_observed = I_0 exp (-t h), where h = airmass). A best-fitting responsivity is chosen by minimizing the reduced chi-squared of the fit.

Fourth, the relative calibration is used to derive opacities. All images are considered, and Beers' Law is applied to every pair of I_observed and airmass. The relative calibration method ensures that (1) substantial calibration uncertainty is not propagated into uncertainty in opacity once sufficient surface data are obtained, and (2) that the processing transfers smoothly from using the laboratory calibration when the first datasets are obtained to using the relative calibration when enough surface data exist.

Data

Each Mars atmospheric opacity data product is structured as two files; a detached PDS label file and a separate data file. Both components are stored as ASCII text. Data within the opacity data file is organized by time with the most recent measurement being appended to the end of the file.

Each Mars atmospheric opacity data product consists of two parts. The first part of the data file contains header information, which includes parameter values used in the opacity computations and column names for the data rows. The second part of the file, starting at line 10 consists of a PDS table object. The table has eight columns and a variable number of rows. There is one row for each opacity measurement. The number of rows in a data product will increase as new measurements of atmospheric opacity are made. Each row is 88 bytes long including the carriage return and line feed characters. All columns are fixed-width as described in the PDS label and are also delimited with commas. Text columns are surrounded by double-quotes and are left-justified. Numeric columns are right-justified.

Software

The ASCII format of the Mars atmospheric opacity data product means that the data can be displayed using a text editor. In addition, the use of the PDS table structure for this data product means the data can be readily imported into spreadsheet and plotting programs.

PDS-labeled tables can be viewed with the program NASAView, developed by the PDS. NASAView is available in versions that run on SUN/SOLARIS, Windows, and LINUX operating systems. NASAView can be obtained from the PDS web site http://pdsproto.jpl.nasa.gov/Distribution/License. There is no charge for NASAView.