: Overview : The MARIE (Martian Radiation Environment Experiment), aboard the 2001 Mars Odysssey spacecraft, was launched on April 7, 2001, and arrived at Mars on October 24, 2001. Data were collected intermittently during the cruise phase, starting in late April and ending in late July. A problem with MARIE's onboard computer occurred in early August, and the instrument was turned off until early March, 2002, after Odyssey's mapping orbit had been established. Data have been collected from that time to the present without major interruption. Routine minor interruptions of up to 36 hours have occurred during the orbital phase; the outages occur when the instrument's data is erased from local storage (after having been downloaded). MARIE is oriented to point in the direction opposite Odyssey's velocity vector. Space radiation is for the most part isotropic, so the orientation of Odyssey is usually not critical and references to external coordinate systems are not a part of the data returned by MARIE. There is one exception to the statement that space radiation is isotropic: During the early stages of solar particle events, there can be directionality in the particle flux. A manuscript containing a detailed description of MARIE is in preparation at the time of this writing (September 2002). : Parameters : The MARIE consists primarily of silicon detectors, of three different types: the A detectors, which are square in cross-section (25.4 mm on a side) and 1 mm in depth; the B detectors, circular, 63.5 mm diameter and 5 mm thick; and the PSDs, or position-sensitive detectors. The PSDs are square double-sided strip detectors with 24 1 mm strips on each side (the strips on one side are orthogonal to those on the other side), and have a thickness of 0.3 mm. There are two A detectors, A1 and A2; sandwiched in between them are PSD1 and PSD2; behind A2, there are the B detectors, B1 through B4. Downstream of B4 is a circular piece of quartz, 10 mm thick, that radiates photons (Cerenkov radiation) generated by the passage of high-velocity particles through it. The Cerenkov photons are reflected by a 45 degree mirror into a photomultiplier tube that sits out of the path of particles that hit the detectors. The MARIE is triggered by a coincidence of hits in detectors A1 and A2. Once triggered, the data acquisition system records 12-bit digitized outputs which are proportional to the energies deposited in the A and B detectors. A two-byte data word is stored for each of these channels. The pulse height from the Cerenkov phototube is similarly digitized in 12 bits and stored. Readout of the PSDs is more complex. Each PSD has two orthogonal sides, referred to as columns and rows. The following description applies to each side of each detector. Onboard hardware analyzes the signals from each of the 24 strips and finds the two largest pulse heights. For each, the pulse height is digitized in 8 bits (256 channels) and stored, along with the strip number. The largest pulse height and position are referred to as ''event1'', the second-largest as ''event2.'' The event2 data are often detector noise not resulting from the passage of a charged particle. Four quantities are stored for each side of each detector, so that a total of sixteen words (thirty-two bytes) of PSD data are stored on each event. The eight-bit digitized pulse heights are referred to as ''magnitudes'' (abbreviated to ''_Mag'' in the list below). The positions (listed as ''_Pos below) are valid only when in the range 1 to 24. The full 72-byte data structure for each event is listed below. Please note well that this structure is not directly reflected in the RDR data files, which are obtained by processing of the REDR files which do have a structure like the one described here: The first nine bytes are instrument and record identifiers, checksums, etc. Bytes 10-15 are the timestamp. Detector data begin at byte 16 and consist of 23 2-byte words. Bytes 62-72 are flags, the first seven of which are unused; the last four are the PSD status registers (one for each side of each PSD). TypeId : Byte; { identifies what kind of record this is } InstID : Byte; { identifies which instrument this is } RunId : Byte; { identifies the current run } RecordId: Byte; { running count of records } len : Word; { 2 bytes - size of actual record on flash } CheckSum: word; { 2 bytes } NumberEvents: byte; { 1 byte total of events per 100 ms. FF max } Time : real; { 6 bytes } Event : EventsArray; { 46 bytes of digitized detector pulse heights } Flags : FlagsArray; { 11 bytes } The following bytes are always the same in every valid record: TypeId (value : 3B hexadecimal), InstID (value : 1), and len (48 hexadecimal, 72 decimal). The first nine bytes are considered housekeeping data and are not kept in the RDR data files. The 46 bytes of detector data are 23 2-byte words, ordered as follows: A1, A2, B1, B2, B3, B4, C1; { for all, valid data in the lowest 12 bits } PSD1_Row_Event1_Mag; { valid data in the lowest 8 bits } PSD1_Row_Event1_Pos; { valid data in the range 1 to 24 } PSD1_Row_Event2_Mag; { valid data in the lowest 8 bits } PSD1_Row_Event2_Pos; { valid data in the range 1 to 24 } PSD1_Col_Event1_Mag; { valid data in the lowest 8 bits } PSD1_Col_Event1_Pos; { valid data in the range 1 to 24 } PSD1_Col_Event2_Mag; { valid data in the lowest 8 bits } PSD1_Col_Event2_Pos; { valid data in the range 1 to 24 } PSD2_Row_Event1_Mag; { valid data in the lowest 8 bits } PSD2_Row_Event1_Pos; { valid data in the range 1 to 24 } PSD2_Row_Event2_Mag; { valid data in the lowest 8 bits } PSD2_Row_Event2_Pos; { valid data in the range 1 to 24 } PSD2_Col_Event1_Mag; { valid data in the lowest 8 bits } PSD2_Col_Event1_Pos; { valid data in the range 1 to 24 } PSD2_Col_Event2_Mag; { valid data in the lowest 8 bits } PSD2_Col_Event2_Pos; { valid data in the range 1 to 24 } The 11 bytes of the flag array are, in order: FlagA1, FlagA2, FlagA3, FlagB1, FlagB2, FlagB3, FlagC1, FlagPSD1_Row, FlagPSD2_Row, FlagPSD1_Col, FlagPSD2_Col Only the last four contain valid data. The RDR data files do not directly reflect the above described record structure, but they are obtained from processing of raw data files which do have this structure. : Processing : These data have been calibrated for the A and B detectors, and converted into physical units rather than digitizer output. In the calibration procedure, cruise data was used to select a sample of events triggered by noise (i.e., with no real particle in the detectors). From these data, the digitizer offsets, also known as ''pedestals,'' were determined. A second sample containing high- energy protons was also culled from the cruise data. Because the energy deposited by a proton in a given depth of silicon is easily calculated, one can obtain the scaling factor needed for each detector channel to convert (after subtracting the pedestal for that channel) digitizer values into deposited energies. The processing involves extraction and calculation of the data values found in the processed data files, including the calculation of the particle incident angles based on the PSD hits. In addition, records or values judged to contain errors are removed or flagged. : Data : These data consist of tabular files (extension .TAB) in ASCII format. Each record consists of 21 columns (23 fields) and is terminated by a carriage return/ line feed (ASCII 13/ASCII 10) line ending. Each record contains two time tags of two fields each; one of these time tags is more heavily corrected than the other. The total length of each record, including line terminator, is 175 bytes. The column structure of each record is summarized below. Column Data Description Type ------ ---- ------------------------------ 1 Character Time (2 fields, year and decimal day, with beginning of Jan. 1 : 1.0) 2 Real Event number (with events numbered sequentially) 3 Real Energy deposited in A1 4 Real Energy deposited in A2 5 Real Energy deposited in B1 6 Real Energy deposited in B2 7 Real Energy deposited in B3 8 Real Energy deposited in B4 9 Real Cerenkov detector pulse height (uncalibrated) 10 Real Row magnitude from PSD1 11 Real Row position from PSD1 12 Real Column magnitude from PSD1 13 Real Column position from PSD1 14 Real Row magnitude from PSD2 15 Real Row position from PSD2 16 Real Column magnitude from PSD2 17 Real Column position from PSD2 18 Real Sum of energies deposited in A1, A2, and B1-B4 19 Real Incident angle of particle (calculated) 20 Character Uncorrected time (same format as column 1, but time tag corrections not applied) 21 Character Error flags for record Deposited energy values in the A and B detectors are given in units of MeV (millions of electron volts). The scale factors and offsets needed to convert digitizer outputs to physically meaningful deposited energy values were determined by the calibration procedure outlined above. Saturation of the A detector electronics is observed to occur at about 70 MeV, and at about 350 MeV for the B detectors, in both cases corresponding to about charge 11 for highly relativistic heavy ions. In principle similar calibration constants could be applied to the PSD magnitudes, but this has not been implemented owing to the generally poor resolution of those data. The signal from the Cerenkov detector is uncalibrated and reported only in units of digitizer channels (0 to 4095 valid). The PSD positions and magnitudes are derived from the full set of reported hits in the raw data, referred to as ''event1'' (which is determined by the strip number that reports the largest-magnitude hit) and ''event2'' (corresponding to the second-largest hit magnitude). Each detector plane (row or column) reports event1 and event2. In the majority of cases, ''event2'' position and magnitude are both reported as 0, which is reasonable because on most triggers only one particle passes through the MARIE detectors. In some cases, event1 may be reported with an unphysical value (0 or > 24), while event2 reports a physically-reasonable value. When this occurs, the processing software outputs the position and magnitude of event2 as the values for that detector plane. The PSD hits are used to calculate the incident angle of the particle. When one or more PSD planes report a value of 0, or greater than 24, the angle cannot be calculated and the output value is set to -5 in order to flag the problem. Valid angles are calculable on a minority of events. When a valid calculation is possible, the formula given below is used, and the value is converted to degrees in the output. Due to the constraints imposed by the A1 * A2 coincidence requirement, only angles less than about 34 degrees are possible. On every trigger, the MARIE clock is polled and the time is stored in a 6-byte data word in the event record. The time is measured in minutes on the Julian calendar, which takes as its starting time 12 hours Universal Time (Greenwich mean noon) on noon of November 24, -4713. In processing, the 6-byte data words are converted to a year and day of year, which includes a fractional day. The day of year is stored, in the processing software, as a 4-byte real number, hence some precision is lost. In the .TAB files, the time is represented by the year, in column 1, along with a real number in column 2 that corresponds to the day of year. The day of year contains 8 digits to the right of the decimal place; the least significant digit equals (approximately) 1 millisecond. Due to rounding errors, and MARIE's ability to acquire data at a fairly high rate, it is possible for consecutive events to appear to occur at the same time, to the precision given here. Invalid values of the year and position columns have been flagged with the value '-1.'. : Ancillary Data : The ancillary data provided with this dataset include files of the following four kinds. These ancillary files are classified as part of the raw reformatted (REDR) dataset. (1) Detector files (file type DET): These are ASCII table files which describe the temperatures of the detectors in the instrument. At the beginning of each file is a table header which names the columns. (2) High Voltage files (file type EXT): These are ASCII table files which describe the maximum voltages in the electronic boards in the instrument. At the beginning of each file is a table header which names the columns. (3) Board files (file type BRD): These are ASCII table files which describe the temperatures of the electronic boards in the instrument. At the beginning of each file is a table header which names the columns. (The BRD files differ from the DET files because each board is only part of a detector and has its own thermistor for temperature monitoring, distinct from the thermistor for the board.) (4) Power files (file type PWR): These are ASCII table files which describe the power consumption of the detectors. At the beginning of each file is a table header which names the columns. : Coordinate System : MARIE's internal coordinate system is defined by the geometry of the detector. The PSDs have their own coordinate system in the sense that each strip (corresponding to a row or a column) is assigned a sequential number from 1 to 24. These values are not referenced to any other coordinate system, as the primary purpose of the PSDs is to calculate the angle at which incident particles traverse the silicon detectors. With valid hits in all rows and columns, the angle THETA with respect to the vertical (normal incidence) can be calculated as follows. We define DELTAr : PSD1_Row_Pos - PSD2_Row_Pos, DELTAc : PSD1_Col_Pos - PSD2_Col_Pos, and d : the distance between PSD1 and PSD2. Then tan(THETA) : [ (DELTAr**2 + DELTAc**2) / d**2 ]**1/2 where ** indicates exponentiation. : Software : A version of the software used for production of calibrated data from raw data files is included in the EXTRAS directory on the archive volume. : Media/Format : Data are archived on recordable DVD (DVD-R) in UDF/ISO 9660 Bridge format with ISO 9660 level 2 compliance in the ISO partition. The data are provided as ASCII tables of time tagged data (see file description above).

: Confidence Level Overview : ------ Review ------ The raw (REDR) data on which these data are based have been reviewed by the instrument team and are of the highest quality that can be generated at this time. The process for production of RDR data has been evaluated by the instrument team and by the PDS. These data will be reviewed collaboratively by the team and by the PDS, and will undergo PDS peer review. --------------------- Data Coverage/Quality --------------------- MARIE was on for parts of the 2001 Mars Odyssey cruise phase, starting in late April 2001 and ending in late July 2001. A problem with the onboard computer occurred in early August, and the instrument was turned off until mid-March, 2002, after Odyssey's mapping orbit had been established. Data have been collected from that time to the present without major interruption. Routine minor interruptions of up to 48 hours have been encountered during the orbital phase; the outages occur when the instrument's data is erased (after having been downloaded). Additional shorter outages of approximately 2 hours duration occur on a daily basis while data are downloaded. A small fraction (typically < 1% of events in a given file) of the time tags in the processed data are out of chronological order. This occurs in the raw (REDR) data and is propagated to the processed data set; it reflects a problem with the MARIE clock that is not understood at this time. As there is no obvious way to correct these errors, they have not been removed or adjusted in the processed data set. In addition, some time tags are corrupted and contained an invalid year. In the processed data, these records have been flagged with a '-1.' in the year field. : Limitations : -------------------- Geometric Acceptance -------------------- MARIE's geometry is optimized for detecting particles coming into the detector (hitting A1 first) in a forward cone of half-angle approximately 30 degrees. The angle is defined with respect to the centerline of the silicon detectors. Particles incident at larger angles may hit one or more detectors, but will not satisfy the requirement of coincident hits in A1 and A2. Forward-going particles encounter minimal (but finite) mass before entering MARIE, and must have residual range sufficient to penetrate as far as A2. As of this writing (May 2002), best estimates of the material in front of MARIE lead to the prediction that the minimum free-space (i.e., before encountering any material) kinetic energy required for a proton to reach A2 is about 30 MeV. For helium-4 ions, the minimum kinetic energy is 30 MeV/nucleon, and for heavier ions, the energy per nucleon increases as a function of the ion's charge and mass. MARIE can also be triggered by particles entering from the rear, provided they are sufficiently energetic to hit both A2 and A1. For a proton, this corresponds to a minimum energy of 72 MeV incident on B4. However, any such particle must pass through a considerable (currently unknown) mass in order to reach B4. (The mass in question is primarily due to Odyssey's oxidizer tank.) Backward-going particles must therefore be far more energetic than their forward-going counterparts in order to trigger MARIE's readout. In the data, backward-going particles can, in some cases, be distinguished by the patterns of energy deposition they leave in the silicon detectors; for particular particle/energy combinations, these patterns may be quite distinct from those of forward-going particles of the same type and energy. In addition, the Cerenkov detector is directional, and no Cerenkov photons will be recorded for backwards-going particles. ------------------------- Saturation of Electronics ------------------------- As of May 2002, all MARIE data have been taken with all silicon detectors operating at full depletion. The gains of the electronics combined with the limitations on input voltage into the analog-to-digital converters (ADCs) - are such that the A channels saturate at approximately 70 MeV energy deposited, while the B channels saturate at about 350 MeV energy deposited. Because deposited energy is a function of the charge (Z) squared of an incident heavy ion, this limits the detectable range of particle charges. In the present configuration, the largest usable signal corresponds to highly-energetic ions of charge 11. For more highly-charged particles, the signals are so large that the digitizer outputs go to full scale, and all resolution is lost. The Galactic Cosmic Radiation contains all elements up to uranium, though ions heavier than iron (charge 26) are rare. At present, and for the foreseeable future, ions with charge 12 and greater are detected by MARIE, but with no resolution. --------------- PSD Performance --------------- The PSD performance for sparsely-ionizing particles such as protons is known to be marginal. The detectors are thin (300 microm), so that signals are small and noise (which falls with increasing detector depth) is relatively large. The onboard algorithm for finding hits returns many 0 values (no hit found), many more values of 1 than is physically reasonable, and in the case of PSD1 columns, many more values of 24 than is physically reasonable. It is likely the case that the edge channels of the PSDs are nosier than those in the middle, and hence are often mistakenly identified by the onboard processing algorithm as having been hit. If the detectors were working optimally, all histograms of row and column positions would show broad distributions peaked in the middle. Only the distribution of hits in the PSD1 rows approximates the expected shape. One possible improvement would be to raise the software thresholds the hit-finding algorithm in the onboard firmware uses. This would result in more ''0'' values (no hit found) but would reduce or eliminate erroneous positions from noise hits. At the time of this writing (Septmber 2002), the instrument team is also investigating the possibility that some non-physical readings are caused by bit flips in the processing firmware. Some of these may be correctable in future processing. ----------------------------- Cerenkov Detector Performance ----------------------------- The Cerenkov detector performance is currently marginal. Signals are far smaller than expected, which suggests that the voltage on the photomultiplier tube is too low or possibly that the alignment of the 45 degree mirror is bad. Attemps to raise the phototube voltage failed due to a bug in the onboard control software. This may be corrected in the future when new control software is uploaded. -------------------------------------- Clock Issues and Time Tag Significance -------------------------------------- Occasionally, MARIE's clock appears to jump backward by a fairly large interval (often on the order of minutes) from one event to the next. The source of this error is unknown. Usually, within the next few events, the clock jumps ahead to where one would expect it to be. Such jumps have been noted to occur up to 25 times in a file of 50,000 event records. As of this writing (September 2002), the instrument team is unable to diagnose or cure the problem. As few events are affected, and most users will likely be binning data in time bins on the order of many minutes or even hours, these do not present a serious problem. A second category of clock problems occurs when MARIE data is being downloaded to the spacecraft's data buffer at midnight. When this happens, the MARIE clock does not increment the day counter, and when data acquisition resumes, the clock is one day behind. It remains in this state until the next data erase sequence (which typically occurs about once every 7 to 10 days, depending on solar activity), at which time MARIE's clock is synchronized to the Odyssey master clock and the problem is fixed (until the next time it occurs). For this reason, deliberate efforts have been made to avoid downloading MARIE data over midnight boundaries. While the problem has not been entirely eliminated, the frequency with which it occurs has decreased. -------------------------------------- Calibration Issues -------------------------------------- There are considerable uncertainties in the calibration constants that are used to convert raw pulse height readings to deposited energies. For the A1 and A2 detectors, we estimate the uncertainties to be +- 20%. For the B detectors, we estimate the uncertainties to be +- 15%. These are uncertainties in the absolute scale of deposited energy; they are not expected to vary significantly within a particular data set, but they may show long-term variations. As of this writing (Dec. 2002), further work to determine these constants with greater precision is in progress, using stopping protons from Solar Particle Events. Periodic monitoring of the constants will be needed to look for, and correct, any drifts in amplifier gains.