Processing Version 188.8.131.52
Last Updated 2007 April 16
|Changes||Internal update of the processing pipeline.|
Currently, the spaced-row charge injection (SCI) is applied to XIS observations by default. However, the SCI completely changes charge transfer inefficiency (CTI). Consequently, current PI values in XIS data with the SCI are not correct, since the PI values are calculated by using the CTI for no-SCI observations in the current data processsing. Current xisrmfgen and xissimarfgen also do not support the SCI. You will be able to analyze SCI data using new FTOOLS and calibration files to be released in June. These new FTOOLS and calibration are implemented in the version 2.0 processing, with which all the data will be reprocessed.
Changes from V184.108.40.206/220.127.116.11/18.104.22.168
22.214.171.124 is a minor update of v1.2.x.x/v1.3.x.x processing. The differences which affect the scientific data are only the CALDB files (XIS microcode and GSO gain-history), and the products' quality is the same as that of 1.2.x.x, and hence any re-processing of 1.2.x.x products will not be performed.The differences from v1.2.x.x processing are:
- Extractor is forced to use DETX/DETY for XIS spectra.
- Bug in creating BURST_ID fixed.
- Cleaned HXD/WEL event files are sorted by time.
- If available, long-term GSO gain history file is used; if not, GSO spectral files are not extracted.
- The STATUS range is set to 0:262143 for both single and multiple XIS input files (it was 0:131072 for the multiple input file case in v.1.2).
Other Notes (unchanged from V126.96.36.199/188.8.131.52/184.108.40.206)
- XIS files:
- Timing mode data are not processed.
- Dead time is not calculated when the burst mode option is used.
- Some of the events might have invalid PI values (PI=4095), when the burst and window options are combined. These events are effectively removed from the subsequent analysis, resulting in a slight reduction in normalization. See Suzaku Memo 2006-20 for details.
- Details of the limitations associated in the data processing are summarized in Suzaku Memo 2006-16.
- The XIS team is preparing background database and tools to be used by Guest Observers for extended source analysis. A temporary version which covers the period 2005 September through 2006 May is available at XIS data analysis page.
- HXD files:
- GSO data in this release are processed by use of the temporary gain correction file which is automatically created in the pipe-line processing, leading to large systematic uncertainties. This sometimes causes improper data screening and the resultant cleaned event files are wrong or sometimes non-existent. To obtain more precise gain correction, including the short-term gain variation, we strongly recommend reprocessing of the HXD data, as soon as the relevant gain history file becomes available in the CALDB. The detailed procedure can be found in the ABC guide and elsewhere.
- The accuracy of non-X-ray background (NXB) model is crucial to obtain accurate scientific results, for a non-imaging hard X-ray instrument like HXD, and the HXD team has been investigating spectral and temporal behaviors of in-orbit detector background. A "background estimator" tool will be released as a part of HEADAS package in the future, but it is still under developement. Until the tool is released, the latest results of the NXB modeling are distributed as "background event files", with 2-4 week delays for PIN and or 4-6 week delays for GSO from the data processing. This delay is because we need housekeeping data from the observation itself, and also we need to reprocess GSO data with long-term gain files. At present, the background subtraction is accurate typically at 5% level, but observers are strongly recommended to check the model validity with a comparison between light-curves of data and background model. The detail of the modeling are described in part in Suzaku Memo 2006-42 (about background reproducibility; other related memos are under preparation).
- Since the dead-time of HXD data is order 4-7% even for background-dominated sources, users should correct it by using the hxddtcor FTOOL. On the other hand, the background event files for the PIN are already dead-time corrected. Those for the GSO are not dead-time corrected.
- Data obtained from 23 March 2006 to 13 May 2006 sometimes suffer additional dead time, clearly seen as dips in raw (i.e. not dead-time corrected) light curve. This is caused by the data transfer saturation within the HXD electronics. The GSO LD was set somewhat lower during this interval. The dips are seen for short periods with high data-rates, e.g. after the SAA and/or in low COR regions of the orbit. In principle, hxddtcor can handle this phenomenon and the spectra should not be affected. Since the issue is yet to be calibrated, however, it is better to check if there is a difference in spectra including and excluding these periods. The phenomenon may also occur even if your data are not within the above interval. The HXD team is now investigating recent trends.
- Common issues:
- Before guide stars are available for the first time (initial 90 minutes or less), both attitude control and ground attitude determination could be less accurate. Users should check images and light curves carefully.
- There is a time-dependent misalignment of 30-60 arcsec, which is synchronized with the orbital period of the satellite (96 min). This is not incorporated to calculate X/Y in this version. A correction tool is under development.
- Spectral Analyis:
- An XIS ARF builder based on ray-tracing simulation (`xissimarfgen') is available (Suzaku Memo 2006-39), while RMF generator is not included in this release. Standard RMF and ARF files can be obtained from CALDB page. These ARF files are limited to point source files for standard pointing positions.
- These standard responses do not include the XIS contamination, which is time, position, and sensor dependent. Users should apply xiscontamicalc to include this effect. Spatial distibution of the contamination will also affect the analysis of diffuse sources, and hence it is recommended to use xissimarfgen. Further details will be documented separately.
- The calibraton of the energy scale is underway. XIS gain is accurate to +-0.2% in the iron band and +-5 eV below 1 keV.
- Effect of the radiation damage is under study now. In addition, inprecise correction for spatially dependent charge transfer inefficienty (CTI) results in line broadening of order 50 eV (in sigma) at Fe K, and a degradation of quantum detection efficiency of about 5% (as of 2006 April).
- XIS calibration accuracy becomes worse in window mode or in 2x2 mode.
- Artificial structure is present in the XIS spectrum in 500-504 ch (1.825-1.84325 keV), which should be discarded from the analysis.
- Local structures may be present in the XIS spectrum with a level of +-10% in 2-8 keV.
- The energy response matrices of HXD-PIN have been verified with the Crab observations, for both of the XIS-nominal and HXD-nominal positions. Results are shown on a separate page. In case of observations performed before 2006 May 24, discrepancies between the data and model are smaller than a few percent in the energy range of 12-40 keV, while they become larger than 3% at >20 keV after the change of one of four PIN-Si bias voltages on 2006 May 29. A newer version of PIN energy response is under development.
- Users should allow 13% uncertainties at the XIS nominal position and 15% uncertainties at the HXD nominal position in HXD/PIN-XIS cross-normalization. For further details, see Suzaku Memo 2006-40.
- Timing Analysis:
- XIS event times are quantized to 8s (see above). Their absolute accuracy is not calibrated yet.
- HXD event times have 61 microsec resolution. Absolute timing accuracy is not yet calibrated to this resolution.
- Barycentric correction software aebarycen has been released. Calibration using this tool is underway.
- Imaging Analysis:
- Exposure map software is not yet included.
If you have any questions concerning Suzaku, visit the Feedback form.