PDS_VERSION_ID = PDS3 RECORD_TYPE = STREAM OBJECT = TEXT PUBLICATION_DATE = 2006-04-20 NOTE = "A lookup table that provides the value and explanation for the QUALITY keyword in the atm*.tab, bol*.tab, obs*.tab, and rad*.tab files." END_OBJECT = TEXT END 1. Overview This document describes the bit contents of the OBS.QUALITY, RAD.QUALITY, BOL.QUALITY, and ATM.QUALITY keywords, including an explanation for each bit and the code definitions for the bit values. The overall observation quality is affected most by spacecraft and instrument motion; it is determined per observation made and stored in the OBS.QUALITY keyword. The quality of the data is evaluated per detector and the results are available in the RAD.QUALITY keyword. Data quality is related to the signal received from the TES instrument and the ground based calibration routines. The quality value stored in BOL.QUALITY, RAD.QUALITY (bits 8-10), and ATM.QUALITY is a credibility flag for the derived thermal inertia and atmospheric products, and is based on the characteristics of the input data and the limitations of the models generating the products. The remainder of this document is divided into three parts: observation quality characteristics, data quality characteristics, and derived products quality characteristics. A brief explanation and the code for the bit values is given for each of the ten quality characteristics listed. The quality bit information can be accessed within a vanilla command using the format QUALITY:field_name where the field names are listed in Table 1 (see SOFTWARE/DOC/USERDOC.TXT for more information). Table 1 - Quality Bit Keywords ATM.QUALITY BIT NO. QUALITY CHARACTERISTIC VANILLA FIELD NAME 1,2 Pressure/Temperature Profile TEMPERATURE_PROFILE_RATING 3,4 Atmospheric Opacity ATMOSPHERIC_OPACITY_RATING BOL.QUALITY BIT NO. QUALITY CHARACTERISTIC VANILLA FIELD NAME 1-3 Thermal Inertia, Bolometric BOLOMETRIC_INERTIA_RATING 4 Bolometer Reference Lamp Anomoly BOLOMETER_LAMP_ANOMALY OBS.QUALITY BIT NO. QUALITY CHARACTERISTIC VANILLA FIELD NAME 1,2 High Gain Antenna Motion HGA_MOTION 3-5 Solar Panel Motion SOLAR_PANEL_MOTION 6 Algor Patch Status ALGOR_PATCH 7 IMC Patch Status IMC_PATCH 8 Momentum Desaturation Status MOMENTUM_DESATURATION 9 Equalization Table Status EQUALIZATION_TABLE RAD.QUALITY BIT NO. QUALITY CHARACTERISTIC VANILLA FIELD NAME 1 Major Phase Inversions MAJOR_PHASE_INVERSION 2 Risk of Algor Phase Inversions ALGOR_RISK 3 Calibration Failure CALIBRATION_FAILURE 4-5 Calibration Issues CALIBRATION_QUALITY 6,7 Spectrometer Noise SPECTROMETER_NOISE 8-10 Thermal Inertia, Spectral SPECTRAL_INERTIA_RATING 11 Detector Mask 1 Problem DETECTOR_MASK_PROBLEM Note that the OBS.QUALITY and RAD.QUALITY keywords may contain up to 32 bits, and the ATM.QUALITY and BOL.QUALITY keywords may contain up to 16 bits. Bits currently not assigned to a particular characteristic are reserved for future use. 2. Observation Quality 2.1 High Gain Antenna Motion The high gain antenna (HGA) was deployed after the end of the aerobraking phase; notable adverse affects due to HGA movement appear in TES mapping data starting at ock 1985. Motion of the HGA induces microphonics in TES and appear as noise in the TES data. Higher rates of motion correspond to higher noise levels in the data. General descriptions of HGA motion can be made according to the spacecraft operation periods. For early mapping orbits, ock 1985 to 3588, the HGA was continually in motion, autotracking throughout most of the orbit with a brief rewind period. For the remainder of "Nominal Mapping" (ock 3589-5784 and 11901-15152), the HGA motion was restricted to occur only during periods of earth contact. During the "Beta-supplement" operations phase (ock 5788-11900 and 15153 to present), the HGA motion remained in autotrack for periods of earth contact and executed complex rewind motions during two periods: straddling the descending equator crossing, and in the southern hemisphere dayside. Appendix A.1 contains information on how the HGA motion data is obtained. Disclaimer: this bit should be used with caution as the information source is spacecraft telemetry which is subject to dropped bytes and can not be completely verified. BIT CODE DEFINITION 0 = HGA motion unknown 1 = HGA not moving 2 = HGA moving at 0.05 degree/second (autotrack motion) 3 = HGA moving at 0.51 degree/second (rewind motion) 2.2 Solar Panel Motion Similar to the situation of the HGA, the motion of either one of the two solar panels induces microphonics in the TES instrument that appear as noise in the data. At the start of the mapping phase the solar panels were continuously in motion, autotracking and rewinding to follow the sun as MGS orbited Mars. Starting at ock 3589, orbit rates and motions were altered to reduce the noise affects on TES; under the new sequence the solar panels only move 3 times per orbit and remain stationary during the interim time periods. This "move and hold" pattern will continue until the end of the mission with the exception of expected periods of power constraints which will require continuous solar panel motion to maintain the health of the spacecraft. The amount of noise present in the data due to solar panel motion is an approximately linear function of the rate of panel motion. The bit values reflect the variety of panel motion rates that may be used. For more information on how the solar panel motion and rates are correlated with individual TES observations, see Appendix A.1. Disclaimer: this bit should be used with caution as the information source is spacecraft telemetry which is subject to dropped bytes and can not be completely verified. BIT CODE DEFINITION 0 = panel motion unknown 1 = panels not moving 2 = panels moving at 0.051 degree/second (non-eclipse, autotrack motion) 3 = panels moving at 0.120 degree/second (during eclipse, prior to ock 3589) 4 = panels moving at 0.240 degree/second (during eclipse, starting ock 3589) 5 = panels moving at 0.400 degree/second (used during aerobraking phases) 6 = panels moving & changing between non-eclipse and eclipse rates 7 = not assigned 2.3 Algor patch status Two patches are simultaneously loaded to correct problems in the TES flight software involving the calculation of the sign of the spectral data and the calculation of the location of the zero path difference (ZPD) in the interferogram. Both of these problems are interconnected and can affect the accuracy of the computed spectra. Better data are produced when the Algor flight software patches are onboard, however some data may still be at risk for problems and can be identified from Data Quality bit 2 (see section 3.2). Algor patch 2A modifies the method employed to calculate the sign of the spectral data by computing the phase for more frequencies, thus improving the phase determined for the output spectra. TES PROM flight software relies upon the symmetry of the interferogram, characteristic to TES-I, to calculated ZPD. The TES-II interferogram is notably asymmetric and another method must be used to calculate ZPD; this alternate calculation is accomplished with Patch 2B. BIT CODE DEFINITION 0 = Algor flight software patch not onboard TES 1 = Algor flight software patch onboard TES 2.4 IMC patch status This bit applies to TES data collected while using Image Motion Compensation (IMC) (see obs*.fmt, IMC_COUNT keyword). The IMC software patch was used to control the direction of steps taken for motion compensation as related to the spacecraft reference frame. For aerobrake orbits, imc moving in the forward direction (bit value 0) will compensate for the spacecraft orbital motion; for mapping orbits, imc moving in the reverse direction (bit value 1) will compensate for the spacecraft orbital motion. BIT CODE DEFINITION 0 = imc moving in forward direction - IMC patch not onboard 1 = imc moving in reverse direction - IMC patch onboard 2.5 Momentum Desaturation status Normal spacecraft operations include routine firing of mono propellant thrusters for a duration of about 3 minutes to adjust the angular momentum of the spacecraft. Any change in spacecraft motion has the potential of introducing noise into the TES data. The amount of noise contributed by momentum desaturation has not been established at this time. Disclaimer: this bit should be used with caution as the information source is spacecraft telemetry which is subject to dropped bytes and can not be completely verified. BIT CODE DEFINITION 0 = autonomous angular momentum desaturation not occurring on spacecraft 1 = autonomous angular momentum desaturation occurring on spacecraft 2.6 Equalization tables status The purpose of the equalization tables is to improve the data compression ratio. These tables should not affect the quality of the data. TES PROM flight software resets the equalization table values to default values after every cold or warm reset. Thus when equalization table edits are loaded for use, the equalization reset patch (2C) is simultaneously loaded. Further information regarding the Equalization Tables is available in "TES Software Specification Document, Instrument Flight Software" [Hughes, SBRC, 1991]. At the time of this writing, the equalization tables have only been used twice: during aerobraking (ock 20-40) and during mapping (ock 8159-8312). Aerobraking: the entropy bits were reset from their default values for Detector Mask 7 to spec_entropy= 4 and reference_det= 2. Compression did not execute as expected, and the equalization tables were removed. Mapping: the entropy bits were reset from their default values for Detector Mask 7 to reference_det= 2. Compression/decompression for full spectral, full spatial resolution spectral data worked as expected; however, this compression/decompression method is incompatible with spectrally and/or spatially masked spectral data. All masked data collected with equalization tables onboard has been deleted as it should not be used for scientific analysis. BIT CODE DEFINITION 0 = equalization tables not onboard TES 1 = equalization tables onboard TES 3. Data Quality 3.1 Major Phase Inversions Spectra with major phase flips or other grossly inaccurate features due to lost bits, incorrect ZPD determination, or excessive "ringing" are identified in this bit. These are major problems with the spectra and possible minor phase flips are not detected here. Appendix A.2.1 contains more information regarding how this bit is identified. BIT CODE DEFINITION 0 = data does not contain major phase inversions 1 = data does contain phase inversions 3.2 Risk of Algor Phase Inversions Spectra with the possibility of inaccurate minor phase flips due to algor problems are assigned a value of 1 in this bit. These flips may not actually be present or recognizable in the calibrated radiance spectra, but careful inspection should be performed before using this data. Appendix A.2.2 contains more information regarding how algor phase inversions are identified. Disclaimer: this bit should be used with caution as the potential for phase inversions has been identified and verified to the best of our ability, but some "low risk" data may actually contain phase inversions. BIT CODE DEFINITION 0 = data at low risk of algor phase inversion 1 = data at high risk of algor phase inversion 3.4 Calibration Issues The spectral calibration algorithm requries space and blackbody reference observations to complete calibration successfully. These reference observations are systematically collected with the data. In the event that either reference observation is unavailable, the RAD.CALIBRATED_RADIANCE fields are filled with "N/A". The RAD.QUALITY:CALIBRATION_FAILURE allows the user to select data which have failed the calibration algorithm (value of 1) due to the lack of the required reference observations. BIT CODE DEFINITION 0 = radiance calibration successful 1 = radiance calibration failed The remaining bits are currently undefined and reserved for future use. 3.5 Spectrometer Noise The value of this bit is a representation of the noise level in the data due to the performance of the spectrometer over time. To completely characterize the noise levels in a particular observation, this bit should be used in conjunction with other quality bits related to noise inducing factors, such as HGA or solar panel motion. Spectrometer noise is calculated from the standard deviation of a 10-ick set of space observations made at least once a day expressly for this purpose. The noise level calculated is applied to all data collected between this and the next 10-ick space observation. Appendix A.3 contains more information regarding how the spectrometer noise is calculated for this bit. The bit value 0 is used for all aerobraking orbits and for mapping orbits where the necessary space observations are not available. BIT CODE DEFINITION 0 = spectrometer noise not calculated 1 = spectrometer noise at nominal levels 2 = spectrometer noise at anomalously high levels 3 = not assigned 3.6 Detector Mask 1 Problem Spectra affected by the onboard detector (spatial) mask problem are assigned a value of 1 in this bit. In March, 2000 the TES Team identified a problem occurring during onboard processing and associated with the use of Detector Mask 1; use of the mask was suspended at that time. This problem is known to affect 0.5% of the surface spectra collected during ock 1723 through 6439. The TES Team strongly suggests that users select only spectra unaffected by this problem, until a method is devised to mathematically correct the problem. BIT CODE DEFINITION 0 = spectrum not affected 1 = spectrum affected by the detector mask 1 problem 3.7 Bolometer Reference Lamp Anomaly Calibration of the visual bolometer data requires regular sampling of one of the two internal reference lamps (see DOCUMENT/PROCESS.PDF Section 3 for details). If the reference lamp looks are unavailable for a significant period of time, the calibration may be adversly affected and calibrated data products should be used with caution. As a result of upload errors, visual bolometer reference lamp looks are unavailable for several hundred orbits collected during July-August, 2001. The reference lamp gap spanning ocks 12064 to 12688 is marked with this bit set to the value of 1; calibrated data products are available, however, they should be used with caution. BIT CODE DEFINITION 0 = reference lamp looks routinely sampled 1 = reference lamp looks missing 4. Derived Products Quality 4.1 Thermal Inertia, Spectral & Bolometric The quality of the derived thermal inertia products is rated and stored in bits 1-3 of the BOL.QUALITY word and bits 8-10 of the RAD.QUALITY word for the bolometric and spectral values, respectively. The ratings are based on percentage uncertainties in the modeled thermal inertia assuming the instrument noise levels stated in the design specifications [Christensen, 1992]. The uncertainty levels are estimated from the partial derivative of log-inertia with respect to temperature, and the ranges associated with each rating are listed in the Bit Code Definition below. Other sources contributing to the total uncertainty of the thermal inertia values are described in DOCUMENT/PROCESS Section 6.0 and in [Mellon et al., 2000]. Thermal inertia values rated as lowest quality (5 to 7) could not be accurately derived due to the reason given in the Bit Code Definition. BIT CODE DEFINITION (for both Spectral & Bolometric bits) 0 = best quality (estimated instrument-noise uncertainty < 1%) 1 = good quality (estimated instrument-noise uncertainty 1-5%) 2 = medium quality (estimated instrument-noise uncertainty 5-20%) 3 = low quality (estimated instrument-noise uncertainty > 20%) 4 = not assigned 5 = lowest quality - observed temperature outside of model-predicted range 6 = lowest quality - no model temperature variation as a function of thermal inertia 7 = lowest quality - thermal inertia value not computed due to lack of necessary data 4.2 Pressure-Temperature Profile The quality of the derived atmospheric temperature profile is rated and stored in bits 1-2 of the ATM.QUALITY word. Currently, only the nadir value ratings of good (0) or not available (3) are in use. These are applied based on the input values of the pressure and temperature boundary conditions: the effective surface temperature (CO2_CONTINUUM_TEMP) must be between 130 and 320 K; the atmospheric temperature (NADIR_TEMPERATURE_PROFILE) for pressure levels greater than 0.1 mbar must be between 100 and 300 K. Other sources contributing to the total uncertainty of the pressure-temperature profile values are described in DOCUMENT/PROCESS, Section 7.1 and in Conrath, et al. [2000]. BIT CODE DEFINITION 0 = nadir values are good 1 = nadir values are questionable (not used) 2 = nadir values are bad (not used) 3 = nadir values are not available 4.3 Atmospheric Opacity The quality of the derived atmospheric opacity value is rated and stored in bits 3-4 of the ATM.QUALITY word. The ratings are based on the thermal contrast between the atmosphere and surface, and the physical meaning of the NADIR_OPACITY results. Opacity values are rated as good (0) if all of the following conditions are met: - thermal contrast is high (CO2_CONTINUUM_TEMP > 220 K) - radiative transfer model opacities and fitted spectral shape opacity values agree (NADIR_OPACITY_RESIDUAL < 0.1) - derived opacity components are realistic (dust: NADIR_OPACITY[1] >= -0.05) (water-ice: NADIR_OPACITY[2] > -0.05) (CO2 hot & isotope bands: 0.05 > NADIR_OPACITY[3] > -0.01) Opacity values are rated as questionable (1) if any of the above conditions are not met. The opacity calculation also depends on the availability of the corresponding atmospheric temperature profile; thus, opacity values are rated as not available (3) if QUALITY:TEMPERATURE_PROFILE_RATING is greater than zero (see above). Other sources contributing to the total uncertainty of the atmospheric opacity values are described in DOCUMENT/PROCESS, Section 7.2 and in Smith, et al. [2000a]. BIT CODE DEFINITION 0 = opacity values are good 1 = opacity values are questionable 2 = opacity values are bad (not used) 3 = opacity values are not available A. Appendices A.1 Determining High Gain Antenna and Solar Panel Motion The high gain antenna (HGA) motion status recorded in the first two bits of the OBS.QUALITY keyword is obtained by combining rate and motion information from the HGA with the time of each TES observation. The HGA rate values were obtained from personal communication with spacecraft engineers (Stuart Spath, Lockheed Martin Astronautics, August 1999) as they are not recorded in the spacecraft telemetry. The HGA motion information is obtained from different sources dependent on how the HGA commands were handled by the spacecraft team. During "Nominal Mapping" operations (ock ~1985-5784 and 11901-15152), the HGA motion is determined from the values encoded in the spacecraft telemetry channels. Channel F-0621 defines the HGA Gimbal Drive Electronics (GDE) elevation motor status as "moving" or "not moving" at specific times, sampled at regular intervals. Channel F-0622 defines the HGA GDE elevation motor direction as "forward", corresponding to autotrack motion, or "reverse", corresponding to rewind motion. Used together, the motion and direction information from these two telemetry channels completely define the status of the HGA. During "Beta-supplement" operations (ocks 5788-11900, 15153-21833, and 25685-30733), the HGA direction is controlled by commands within the Spacecraft Command Files which are uploaded to the spacecraft every 3-4 days. The command code SAHACE begins HGA autotrack motion; the command code SALHTA begins a specific HGA rewind maneuver. By combining the time sequential command codes (for direction) with the spacecraft telemetry from Channel F-0621 (for motion) the status of the HGA is again completely defined. The solar panel motion status is determined in a similar manner. The motion of each solar panel, SAM and SAP, is defined independently in separate spacecraft telemetry channels, F-0801 and F-0821 respectively. In each channel the elevation motor status is given as "moving" or "not moving" at specific times. Again the rate of solar panel motion is not recorded in the spacecraft telemetry and was obtained through communication with the spacecraft engineers. The solar panel rate varies over the course of the mission, and also throughout the course of an orbit. During a single orbit, two rates are used: a faster rate during solar eclipse periods, and a slower rate for autotrack motion during non-eclipse periods. From the point of view of the spacecraft, the time of solar eclipse varies by orbit and must be obtained from the heliocentric surface occultation beginning (SOCCSB) and end (SOCCSE) time entries in the Orbit Propagation and Timing Geometry (OPTG) file. To fully assign the solar panel bit value in the quality word, the telemetry motion information, the known rates, and the solar eclipse times must be combined with each TES observation and associated time. Because the solar panels communicate telemetry only at specific time intervals which may or may not correspond with each TES observation time, some logical interpolation was applied to determine the value of these bits. For example, TES observations obtained at a time that falls between a telemetry record showing motion during an eclipse period and a record showing motion during a non-eclipse period (or vise versa) would be tagged with the bit value 6 corresponding to panel motion during a transition period. A.2 Determining Phase Inversions A.2.1 Major phase inversions and other grossly inaccurate spectra An algorithm detects major phase flips or other grossly inaccurate features due to lost bits, incorrect ZPD determination, or excessive "ringing". These are spectral problems that are clearly identified when the spectrum is plotted, but may not be noticed otherwise. The algorithm checks for these problem spectra using two methods: specific thresholds and derivatives. The threshold checks that uncalibrated radiance values are within specific thresholds in several wavelength regions: 200-220cm-1 (value range -10 to 3); 645-680cm-1 (value range -80 to 1); and 1610-1650cm-1 (value range -12 to 7). If any spectral channel lies outside this range of values, the spectrum is determined to be bad and a value of 1 is assigned. The derivative check takes the derivative of the spectrum from 200-530cm-1 and 800-1200cm-1. If the absolute value of any derivative throughout this range is >15, then the spectrum is assigned a value of 1 indicating a problem with the spectrum. A.2.2 Algor phase inversions Algor phase inversions are due to low temperature contrast between the sensor and the target, and it occurs because the phase of the spectrum is interpolated between a number of points in the spectrum. In spectral regions where the measured voltage is near 0, it becomes impossible to interpolate this value exactly. This only occurs at the shorter wavelengths (< 850 cm-1) and has not been observed at longer wavelengths. The algorithm that checks for possible minor phase flips due to algor problems is fairly strait forward. The uncalibrated radiance spectrum is scanned between 850 and 1400 cm-1 for values with an absolute value of less than 1. This is an arbitrary threshold where the phase flips have been known to occur. Where the entire range is either above 1 or below -1, the phase flips are assumed to not be present and the spectrum is assigned a quality value of 0. If any spectral sample in the spectral range inspected is within these bounds, then the spectrum is assigned a bit value of 1 indicating that the likelihood of minor phase flips is probable. A.3 Determining Spectrometer Noise The spectrometer noise recorded here is a representation of the results from a study to monitor the health of the instrument over the course of the mission. For this study, 10-ick space observations are routinely collected at least once a day. The raw radiance for the 10-ick set is averaged together and the standard deviation is calculated from 3 selected wavelength ranges: 300-400 cm-1, 900-1000 cm-1, and 1500-1600 cm-1. Finally, the average value of the standard deviation in these three ranges is used to define "nominal" or "anomalously high" levels of spectrometer noise. For single length scans, standard deviations of 0.00 to 0.28 are considered nominal spectrometer noise; for double length scans, standard deviations of 0.00 to 0.40 are considered nominal spectrometer noise. The space observations used in this study are collected during periods when neither the high gain antenna nor solar panels are moving, since both induce increased noise in the spectrometer. A strategy to specifically target data collection in periods of non-motion for spectrometer noise analysis has been in effect since ock 3589; before this time space observations collected during periods of antenna or panel motion may have been used if no others were available.