Release Notes for Level 2 SIS processing C.M.S. Cohen Version 2.1 June 6, 2008 cmsc The largest change to the level 2 analysis is the addition of more elements. We have added intensities for Na, Al, Ar, Ca, and Ni. It should be noted that for Na (and potentially Al) the intensities during large SEP events maybe more uncertain due to spill over from the neighboring abundant elements. Although we have not corrected the Na intensities on a case by case basis (which would have to be done by hand), the uncertainty in the absolute intensities should be <20% during high rate periods. It was determined that the detection efficiency for small pulse heights in one of the matrix detectors of the B telescope has fallen significantly since 2005. This primarily affects long range He. We have reprocess the data from 4 January 2005 onwards (Bartels rotations 2340 and beyond) to make use of only data from telescope A for calculating the He intensities. This effect is not a concern for the heavier elements so that processing has remained unchanged. A few minor changes in the analysis code have been made. One deals with the handling of noisy strips. These strips are ignored when a better alternative exists for the analysis. Unfortunately, the code was not set up to handle a list of noisy strips that was not in strip number order or to account for reactivated off strips. The code has been updated to correctly handle these situations. A second item is that the code has been updated to automatically calculate the revised geometry factor based on the current listing of off strips. Version 2.0 April 18, 2007 cmsc Major upgrade (including a new major program) of the level 2 processing has been done. The previous level 2 program (level2_flux.pro) is no longer being used. The required program and subroutines are given below. The largest change in the processing will be evident in the range 0 intensities. This is primarily due to a new analysis check which removes events that penetrated deeper in the instrument but were incorrectly tagged as range 0 events due to incomplete collection of the deeper range pulse heights. The upgraded analysis will result in lower range 0 intensities in general. Another large change is evident in the range 7 He intensities. This is due to a reanalysis of the trigger levels required for deep range He. The upgraded analysis will result in higher range 7 He intensities in some time periods (particularly quiet times). A listing of the current analysis checks and cuts is given below. For reasons of brevity the descriptions are terse, but feel free to contact Christina Cohen (cohen@srl.caltech.edu) for a more complete description/ explanation. We remind users that effects due to low livetime conditions are still present in the data. When particle intensities are very high (generally when the >10 MeV Proton rate exceeds ~2000 (cm^2-sr-sec)^-1) large corrections for low instrument livetime and pileup are somewhat uncertain. This may result in the reported absolute intensities to be low (by an amount that is undetermined at the moment, but probably less than a factor of 10). Since all the Z >= 6 intensity calculations involve the same livetime, the relative intensities of the heavy ions should be reasonably accurate even during these high rate periods. Users of daily and Bartel rotation averaged files should be aware of the limitations of calculating uncertainties based on the square root of the reported number of particles (i.e., counts). For averages that include time periods of very low livetime, the average intensities may be dominated by counts accumulated during the those low livetime portions. Since the counts are typically low during these periods, the uncertainty of the corresponding intensity is high. This should be taken into account when calculating uncertainties for the full time-average period, but will not be if the user calculates uncertainties based on sqrt(counts). We encourage users to examine the intensities and counts for a higher time resolution product during such instances to correctly assess the uncertainties. Current analysis checks/cuts (new items are indicated by *): - removes stim events - corrects geometry factor for 15 off strips (through 2002 1/21) - includes a patch for noisy strips (ignores strips considered to be noisy - time dependent) - removes out of geometry events - requires He events to be in buffers <= 60 - requires HIZ events to be in buffers > 60 - corrects for priority system and livetime -*corrects range based on low pha energies before stopping detector (ranges 3-7, not for He) -*cuts based on time to peak -*uses better gains and offsets for matrix energy calculations -*adds neighboring matrix strips together -*uses better matrix thickness maps -*cuts based on consistency between ground and high voltage sides of detectors -*better cuts based on Z consistency -*He R7 uses T4 for dE in one of the Z calculations rather than T1 -*cut based on coincidence with He for range 0 HIZ events -*cut based on consistency between energy deposited in stack and matrix detectors -*cut requiring range 0 HIZ events to have a minimum energy deposited in M2 -*cut requiring trajectory to land inside both matrix detectors - buffer rate spike filter applied - changes range 0 He data by factor of 1.6 and zeros out range 1 He data prior to 1998 1/28 22:40 due to thresholds - livetime includes correction for timing window Processing prior to ASC processing: Primary Program: make_level2.pro Language: IDL Required Subroutines: read_calcfile, read_calcfile_HHe, fluxcalc, various standard ACEware routines Run on files: *.HIZ and *.HHe, created by make_calcfiles* which runs read_datafile_se1_pro these files are stored in 10 day increments on /home/idunn1/cohen/ Program: level2_flux.pro Language: IDL Required Subroutines: pha_livetime.pro, various ACEware routines Version 1.6 November 17, 2005 cmsc Code has been added to the level 2 processing to ignore signals from known noisy strips on the matrix detectors. This should improve the trajectory calculations and subsequently the elemental resolution leading to more accurate intensities. The effect was most noticable for Bartel rotations 2349 and 2350. These have been reprocessed. Version 1.5 January 22, 2002 cmsc Data from several big events (most notably the July 14, 2000 and November 9, 2000 events) have been added to the level 2 dataset. During the highest-rate portions of these events (see below), large corrections are necessary for pileup and instrument livetime which are somewhat uncertain. The resulting reported absolute intensities near the peaks of these events are probably low and for the singles rates (such as the >10 MeV Proton rate) can be uncertain by as much as a factor of 10. It is expected that the heavy ion (Z >= 6) intensities will be uncertain by a smaller factor but the precise amount is unknown at this time. Since all Z >= 6 intensity calculations involve the same livetime, the relative intensities during these events should be reasonably accurate. In general, these livetime issues are a concern when the >10 MeV Proton rate exceeds ~2000 (cm^2-sr-sec)^-1. High-rate, suspect periods: 2000 7/14 12:00 - 7/14 16:00 (when thresholds were changed) 2000 11/9 00:00 - 11/10 00:00 2001 11/5 16:00 - 11/6 08:00 2001 11/23 12:00 - 11/24 16:00 Users are further cautioned that during the November 9, 2000 event there was a noisy strip on SIS that drove the livetime down precipitously. This resulted in a substantially reduced number of good events being telemetered. Users will see the effects of this most strongly in the scant Fe counts during the peak of this event. Version 1.5 January 1, 2001 cmsc A new formula for the event livetime was determined that is more accurate than the simple instrument reported livetime during time periods when the livetime is very low. This correction results in more accurate intensities of all heavy ions during several big events. Bartel rotations during which these events occurred were reprocessed: 2283 2275 2257 2256 2254 2253 2249 2243 Version 1.3 October 12, 1999 cmsc It was determined that long range He often may not trigger the T1 detector causing the event to be thrown out by the current z consistency cuts. This caused range 7 He fluxes to be artificially low. One of the two consistency cuts for He was eliminated and resulted in an approximate increase in the range 7 flux by a factor of 2. It has also been observed that in time periods during which the detector thresholds were changed (during 3 large solar events) the low Z, high range fluxes are artifically low. This is probably due to failure of events to trigger the T1 detector and and being thrown out by one of the consistency checks. No correction has been made and the user is cautioned about low fluxes in the high ranges for low Z elements appearing during the following time periods: 97/11/07 16:00 - 97/11/11 16:30 98/04/21 15:10 - 98/04/25 16:05 98/08/25 16:40 - 98/08/27 16:55 Quiet time selection is being made by the ACE Science Center. Examination of the quiet time data indicates there the low range He data are not entirely quiet. Since tightening the quiet time criterion may eliminate quiet time heavy ion data, we have chosen not to change the criterion but suggest that users be aware that quiet time low range He may still contain some solar activity. Version 1.2 September 10, 1999 cmsc Better z consistency cuts were determined. This will increase the flux levels at the high ranges (6 and 7) by about ~10% for CNO, by about ~5% for Ne, Mg and less than 1% for Si-Fe. Currently we suggest a 5% systematic error be added to all the flux uncertainties. In comparing the range 7 data with range 2 data from CRIS we have found an ~10% discrepancy. We will continue our efforts to determine the cause of the discrepancy and correct it, but currently take it as a measure of how accurately we know the absolute flux levels. Headers on the files were changed to tabulate the new Z consistency cuts (which are now a function of Z, range and which Z estimates are being compared - there are 2 comparisons being made). See program for a complete listing of the zcut boundaries. ** All data were reprocessed as of September 10, 1999 ** Version 1.1 August 13, 1999 cmsc Better nominal geometry factors were obtained. The following are now used: GF for He: GF for C-Fe: Range 0 = 42.8 Range 0 = 42.8 Range 1 = 42.8 Range 1 = 42.8 Range 2 = 41.2 Range 2 = 41.4 Range 3 = 41.2 Range 3 = 41.2 Range 4 = 39.0 Range 4 = 39.0 Range 5 = 37.6 Range 5 = 37.6 Range 6 = 32.2 Range 6 = 32.2 Range 7 = 24.6 Range 7 = 24.6 Note: the entry on Jun 29, 1999 regarding actual geometry factors used still applies, i.e., these nominal geometry factors listed above are reduced to account for disabled matrix strips when fluxes are calculated. Instead of providing errors in the flux for each range we are now providing the raw number of counts in each range. ** All data were reprocessed as of August 13, 1999 ** August 5, 1999 cmsc On Jan 28, 1998 at 22:40 UT the matrix thresholds on SIS were lowered. This resulted in a change in the SIS efficiency in measuring He fluxes which predominantly affects the range 0 and 1 data. In order to make the He data prior to the threshold change available and useful to the community we have made the following modifications to the data. For data prior to Jan 28, 1998 22:40 UT -- He range 1 is not given, the data in the files is fill data. He range 0 has been increased by a factor of 1.6 which corresponds to our best estimate of the efficiency reduction related to the higher matrix thresholds The quiet time flag is calculated using the average of the He range 0 and 2 fluxes (rather than the average of the He range 0, 1, and 2 fluxes) For data after Jan 28, 1998 22:40 UT -- He fluxes for all ranges are given. The quiet time flag is based on the average of the He range 0, 1, and 2 fluxes. On Feb 10, 1998 at 18:26 UT the H and He timers were altered which affects the amount of H and He livetime. When examining the He data at maximum resolution (256 seconds), the difference in livetime results in a factor of 2.4 increase in the minimum detectable He fluxes (i.e., the single count level) for all ranges. When averaged over a longer time interval (1 hour or greater) it becomes clear that the actual measured flux value is unchanged by the timer command. ** He data for the affected rotations (2241-2246) were reprocessed as of August 6, 1999 ** Version 1.0 June 29, 1999 cmsc A filter to search for and eliminate spikes in the rate data (used to convert to flux) was added to the pha_livetime.pro subroutine. The algorithm for identifying spike is to compare each point to its 2 non-zero neighbors (one on each side) and if the point is greater (less) than a factor of 50 higher (lower) than both neighboring points the point is labeled as a spike. The corresponding flux data for the entire 256 second period is zeroed out (regardless of Z). A filter to identify spikes in the calculated flux due to HIZ Z events tagged with a buffer ID corresponding to He or H was added to the level2_flux.pro program. The bad data were identified by requiring all events with Z > 5 to have buffer ids < 60. The geometry factors were altered to reflect the fact that at launch 2 strips (of 64) were off on detector M1B. After 1997 Dec 8 (day 342) 17:42 UT, 4 additional strips on M1B were turned off. The nominal geometry factors assumed identical telescopes (i.e., gf(tel A) = gf(tel B)). The correction made was to multiply the nominal geometry factor of telescope B by a factor of 62/64 from launch to day 342 17:42 UT and by a factor of 58/64 for times after day 342 17:42 UT. ** All data were reprocessed as of August 3, 1999 ** Version 0.1 Original program outputs 256-second data for a given Z range (corresponding to one element) for all 8 instrument ranges. 1 file is created for each element. Data are in flux units. Uncertainties are fractional uncertainties and based on statistical uncertainties only (i.e., 1/sqrt(N)). Currently range 8 flux and uncertainties are set to 0. Nominal geometry factors and energy intervals are used: GF for He: GF for C-Fe: Range 0 = 41.4 Range 0 = 38.4 Range 1 = 41.4 Range 1 = 38.4 Range 2 = 41.4 Range 2 = 37.5 Range 3 = 40.3 Range 3 = 36.7 Range 4 = 38.4 Range 4 = 34.4 Range 5 = 37.0 Range 5 = 33.1 Range 6 = 31.8 Range 6 = 27.2 Range 7 = 24.5 Range 7 = 19.4 (See program for nominal energy intervals) Very loose z consistency cuts are made to the data. Z1/Z0 and Z2/Z0 are required to be 1.0 +- 0.03, where Z0 is Z calculated using dE = sum of all previous detectors, E' = stopping detector, Z1 is Z calculated using dE = M1 detector, E' = sum of all following detectors, and Z2 is Z calculated using dE = next to last detector, E' = stopping detector.