Scientific Validation
Contents
1. o iS 0 RA CAR deg RA CAR deg RA CAR deg Fig x2 spiskymax images for different source energies source distance 125 m event types singles and doubles and triples Sources are Am left 59 keV on axis ad Ox 662 keV center at 2 aspect see below for variations across the field of view and Na 2754 keV right on axis o RA CAR fdeg RA CAR fdeg RA CAR fdeg Fig x3 Co spiskymax images source distance 125 m under different viewing conditions Source aspect rotation by 30 left and viewed with 11 pointings with different aspects center and expanded view right From BLC calibration measurements imaging analyses indicate the accuracy and width of the point source response Note that the source beam divergence is 8 hence the source is not at infinity as cosmic sources will be From a single source exposure the achieved angular location precision at 1173 keV is better than 12 figure x1 the width of the residual extent in the spiskymax image SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 11 of 21 RA CAR deg RA CAR deg RA CAR fdeg Fig x4 6Co spiskymax image sources distance 125 m event types singles and doubles Here available source aspects were combined such as to effectively emulate a multi point dither pattern The left image shows a single source viewed by 11 pointings The center image shows two sources se2. SPIDEAD Proper uncontaminated data intervals are derived with SPIGTI see performance analysis Imaging responses and background models are tailored to the analysis data sets with ISSW modules SPIBOUNDS define binning constraints in general terms BIN I define the binning used in analysis IMG I prepare the response matrices in appropriate binning and BGK I prepare relevant background model The measured data themselves are binned into histograms for science analysis using ISSW module SPIHIST Utilities for interfacing to previous knowledge about the gamma ray sky are provided for diffuse emission skymap convolution ISSW module SKYCNV with preparation of the sky parameters in module GENSKY and for source calalogues ISSW module CAT l ISSW tools for astrophysical analyses in the areas of imaging spectra timing and model fitting The most general and theoretically also most sensitive analysis would make use of the full data in unbinned form to retain measured resolutions and use the instrument response in full spectral and spatial detail to deconvolve the appearance of the sky Response information would be too complex and big to be handled however so that compromises are necessary For imaging analyses one assumes the separation in wide energy bands is adequate to not distort results from adjacent energy bands For spectral analyses one either attempts to separate a source direction through a first round of imaging which
3. document we review the performance that can be expected from the realistic mission situation with its exposure patterns and after standard analysis tools have been applied to generate results Energy calibration and gain correction tbd Here we would like to have fitted energies from preprocessed event data gain corrected with standard tools after fitting the predefined set of background lines with the standard calibration analysis tools a table with the achieved energy scale accuracy versus energy is the goal CESR Dead time and effective observation time tbd Difficult to test these without real data AWS simprep tools review Use of BLC data after their embedding into simulated obs pattern Goal is to demonstrate that identical source intensities are recovered under different deadtime countrate conditions and with different observation patterns MPE and CEA SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 10 of 21 Angular resolution source separation power and location accuracy Fig x1 6Co spiskymax image source distance 125 m event types singles and doubles and triples No ghost images appear left the expanded view on the right side shows the accuracy and width The middle image was derived from all event types while the righthand image was derived from double and triple events only skymax_image_1 fits_2 skymax_image_l fits_2 skymax_image_1 fits_2
4. formatting to provide imaging response matrices IRF for the SPIROS SKYMAAX DIFIT imaging analysis modules and spectral response matrices ARF RMF for XSPEC spectral fitting ISSW tools for background analysis and modelling Instrumental background is large for any instrument in the MeV regime from cosmic ray activation of spacecraft material Basic standard spectral analysis software e g from CERN packages as part of the ISDC ROOT scripting language but also modules have been provided as part of the ISSW such as SPILINE or GASPAN is used to identify characteristic spectral lines and thus explore the background types Fit results must be digested interactively by instrument scientists to derive background parameters no direct interface to background modelling is foreseen see below For detailed Monte Carlo simulations of background both the MGEANT package used also for the response simulation and the TIMMS implementation of GEANT with enhancements are available External particle radiation environments are specified by the user the detailed mass models and response functions included in these packages then generate the simulated Ge detector event messages as expected from such background Studies performed at CESR Pierre Jean and CEA Nene Diallo constitute a baseline for SPI instrumental background More must be learned during the mission from comparisons of background explorations with simulations Background handling in th
5. no warnings os_pdefgen needs to be run to get the AT Tltude history file Parameters dither pattern definitions are redundant and many do not actually work Using the standard GCDE code 9 produces an error of some odd number code 18110 and complains about improper specification Trying Galactic coord s also comes with strange error messages about inadequate specification of Z1 which is none of the parameters In any case the generated image always is calculated in RA Dec one cannot change that to Galactic So generating the desired observations composed of sets of standard 5x5 dithers or the GPS GCDE was not possible For the Cygnus simulations chose the XRA XDEC parameters for the fov axis standard 5x5 dither but no idea about settings of the position angle and z coord spec s leave them at default Crash SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 7 of 21 segmentation fault after understood messages which don t mean anything to me caused by user spec of RA in degrees while the program expected hh mm ss the parameter instructions could be clearer here spisimprep runs ok generates GTI deadtime etc Apparently the GSFC simulated aux files do not quite match these file formats however although the ATTI files look identical when plotted spiskymax refuses to accept the input files from the GSFC cygnus simulation why No idea all combinations of available files tried
6. spiback Error No 2004 finally was found to mean that a detector event spectra file evts_det_spec fits was missing this is an empty dummy according to AWS but must exist The SPIBACK help file is inadequate it plainly refers to the ICD for an explanation of the parameters this is not helpful at all Otherwise spiback works Apparently however the present version only can use pre set bgd levels no connection scaling with aux parameters is supported spiskycnv Needs responses IRF as stored in Andy Strongs private directories where would find the ones to be used now chose that SPISKYCNV interprets the gensky image in Galactic coordinates see above This concept that the map can be interpreted in different coordinate systems and hence can mean very different things should be re considered invitations for errors It is not prominently stated that HERE the gensky map and the spiback background are combined to produce the simulated dataset so beware that spiback has been properly applied before spiskycnv is run General Remark on Test Data Provision at this Level In ongoing work GSFC have produced new simulated data sets using our SPI model integrated onto TIMM3 4 to assess the effects of a bright off axis source passing through the IBIS i e the SPIBIS instrument These data can be treated separately or combined with the Cygnus region data to assess the total effects GSFC are also working on a set of mono energetic
7. utility perhaps incorporating a GUI interface is needed Sspi_gain_cor There are few programable inputs to this program so its usability is straight forward and not much to assess We did compare the results of spi_gain_corr directly against spihisto Toulouse version for various runs and several event types and find perfect agreement in all instances We had some confusion over the cleaned PSD multiple events in terms of the definition of their selection criteria and bookeeping impact on other multiples This confusion was nominally resolved by reading between the lines in the ICD Performance is a concern Even for single science window analyses this was evident spidead Straight forward to use no specific problems to report However at the moment seems to apply a somewhat arbitrary scale factor independent of detector event SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 6 of 21 type At some point an attempt at a physics oriented deadtime calculation needs to be implemented spipoint Straight forward to use no specific problems from our perspective spigti We had early problems with time windowing as pertains to spihist event selection but this is now resolved spibounds Straight forward to use no problems to report Offers flexibiliy in binning schemes to support scenarios unique to gamma ray line studies line region s can be densely sampled while continuum regions c
8. 0 0032 0 0271 0 0053 0 8000 0 0028 2 1 600 285 7 0 0016 0 0322 0 0077 0 4016 0 0020 4 1 606 202 1 0 0025 0 0133 0 0104 0 1985 0 0014 8 1 588 142 4 0 0030 0 0812 0 0164 0 1019 0 0010 16 1 630 101 3 0 0129 0 0293 0 0162 0 0999 0 0010 16 1 599 100 5 0 0064 0 0148 0 0154 0 1011 0 0010 16 1 618 103 8 0 0157 0 0417 0 0163 0 1001 0 0010 16 1 602 99 0 0 0004 0 0312 0 0141 0 1023 0 0010 16 1 637 102 6 0 0030 0 0438 0 0159 0 1009 0 0010 16 1 615 102 0 0 0224 0 0222 0 0223 0 0504 0 0007 32 1 613 71 8 mean 0 0019 0 0335 0 0131 1 610 rms 0 0098 0 0184 0 014 Table 1 Results from Spiros on source location and recovered flux taking BLC data for Cs and reducing the source significance by successively reducing the number of photons by taking random selections SPI ISDAG 14 05 02 issw scival_ V1 doc SPI ISSW Science Validation Report Page 12 of 21 Devector pteeler Dereace pbeeler X0 Joo Joo iso 10 Dd 4 D D Joo Dereace pteeler Dereace pteelei Fig X5 Images of simulated celestial sources in the Virgo region as observed with realistic dither pattens and background A spiskymax image for the 3C273 and 3C279 quasars simulated with 1 0 sec exposure in a Standard 5x5 dither pattern with 2 separation is shown on the left side in the 400 1000 keV energy band The righthand image shows a spiros image at low energies 50 150 keV from a simulation of 3C273 alone with realistic background and an 11x11 point dither
9. IHISTO line fitting ISSW module SPILINE then determines the gain correction factors ISSW module SPICALI Similarly from Performance Analysis a set of assessed PSD calibration libraries is maintained from which the PSD classification criteria are derived Both the gain correction and PSD classifications are applied ISSW module SPICOR when the events are read used for science analysis e g when binned into spectra for later analysis ISSW module SPIHIST The instrument response of SPI is determined solely from Monte Carlo simulations of the physics interactions within instrument and detectors The MGEANT software package is used to handle the physics details and produce physical interaction events The SPI response is determined as a function of incidence direction and energy The different variabilities of response aspects with angle and energy is accounted for by decomposition into different matrices L D one addressing the detailed energy response of detectors the other addressing the attenuation of gamma rays from all incidence directions on their way to the Ge detectors The ISSW module RSPGEN composes these matrices to assemble a response function representation as needed for different purposes i e for spectral analysis or for imaging analysis The basic response matrices L and D are provided to ISDC by GSFC of the SPI Team The RSPGEN module uses these and makes the necessary interpolations instrumental preformance adjustments and
10. SPI ISSW Science Validation Report Page 1 of 21 SCIENTIFIC VALIDATION OF SPI INSTRUMENT SPECIFIC SOFTWARE Document ID SPI DAG MPE ROD 20020514 by SPI ISDAG Roland Diehl lssue 1 14 May 2002 1 About this Document Scope Here we assemble the plans methods and results of the scientific validation of the SPI instrument specific software as embedded in the software system for INTEGRAL data analysis at ISDC We focus on instrument system level here and refer to supplementary scientific validation logs amp reports and integration reports per ISSW tool for further details Document History e 26 Mar 2002 Draft 1 planning created from ISDAG MM and ISSW documents e 06 May 2002 Draft 2 including reports from data prep and imaging for ISDC Mtg e 14 May 2002 Issue 1 revised according to ISDC Mtg comments for distribution 2 ISDAG s Software Validation Plan Validation Goals Comments It was felt that starting to play with existing tools would sufficiently guide the tester into the test objectives and a sort of referee report on a tool area would be appropriate Even though this approach is biased by existing software it was felt that an initial step of generating a test validation plan with an inventory of test questions and success criteria could be spared Validation of ISDC Tools in general is understood to exercise a tool in a near realistic environment on near realistic problems in order to assess the adequacy of
11. W tools for instrument performance analysis ISSW tools for near realtime analysis of bursts and transients ISSW tools for instrument calibration and response determination ISSW tools for background analysis and modelling ISSW tools for preparation of data for astrophysical analysis oS oe eS SS ISSW tools for astrophysical analyses in the areas of imaging spectra timing and model fitting Preprocessing and routine monitoring parts of this system is ISDC specific and response calibration and performance analysis parts are specific to project related sites and the ISDC while science analysis parts of this software may be distributed to the community with the data ISSW tools for data decomposition and preprocessing The SPl specific telemetry aspects have been encoded into ISDC s Preprocessing software based on the SPI Science Data Format Description ISSW tools for instrument performance analysis The SPI scientific performance is monitored with ISDC s OSM observation status monitoring this is a root based package with an automatic and an interactive implementation Interactively a variety of user specific displays can be built and saved for later usage This interactive OSM is also used by SPI experts for their deep science performance analysis work at ISDC and remotely Specific OSM displays are provided for the SPI hardware subsystems such as the Anticoincidence System ACS For subsequent science analysis
12. aging mode for a number of scenarios Cygnus region point sources mono energetic point sources at variosu energies offaxis angles For Cygnus the fainter softer sources were not well determined even though there was no background included had difficulty performing followup runs in spectral mode This seems to be a DAL file management problem rather than a spiros specific problem Performance and memory management are a concern For a small 9 energies set of IRFs the image reconstrution for a single point source with no background took about 10 s of minutes typically The new just delivered set of IRFs cover 50 energies and we need in the near future to include background determination in the image reconstruction tests spidiffit no third party assessments yet Spectral Analyses spiros no third party assessment reports yet spiskymax no third party assessment reports yet spidiffit no third party assessment reports yet SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 9 of 21 4 SPI Science Performance Parameters Here we assemble the SPI instrument performance parameters as obtained through analyses of calibration and simulation data with the ISDC ISSW tools We attempt to follow the parameters as identified in the SPI Science Performance Report Yet in that document the emphasis is placed on the instrinsic hardware detector performance parameters while in this
13. al resolution SPI instrument in many cases therefore spectral information also is derived from imaging analysis tools Timing Analysis No special ISSW has been provided for source timing analysis XCHRONOS application on lists of event time tags is the ISDC standard A special imaging preselection analysis as described for spectral analysis is being worked on SPI ISDAG 14 05 02 issw scival_V1 doc
14. aspects MPE Roland Diehl Imaging tools Source parameter and spectral aspects MPE Andy Strong Imaging tools diffuse sources aspects UBham amp CESR Gerry Skinner In practice validations of different types were made by these and several other people addressing thd e tbd e tbd SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 4 of 21 3 Validation Report General Findings Here we list observations which apply to the system as a whole or to several of the tools 1 The use of the ISDC system has a fairly high entry threshold The user s environment must be carefully prepared and tested such that all environment variables are properly set and the access to software and data repositories works There is no guidance on tool names at the beginner s level names of tools must be known in advance Familiarity with Unix features such as sophisticated grep and emacs and with utility features such as the fv display options and tailoring are essential It would be advisable to add an introductory primer brief reminder for the non developers where one finds the tricks how to know which tools exist and what they do the general intro format and use of par files See SPIROS SUM 3 1 and its editing with the available editors some tricks for where files are expected and produced and how to efficiently organize this All this is obvious for insiders and familiar to regular users but
15. ated with these in the proper data group seems a realistic intermediate method adequate for validation of the analysis tools For the validation exercise astronomical test cases are Crab source with a line feature and powerlaw spectrum EBOUNDS POINTING and DETESPECTRA datasets as FITS files spectra in cts bin Crab like power law source with a superimposed 440 keV line background spectral form resembling Jean et al 1997 superimposed but scaled for exposure of 34 5 ks first provided Dec 2000 updated Feb 2002 Cygnus region with set of sources with different spectra and a diffuse component simplified spatial pattern 4 point sources including 2 black hole Cyg X 1 and V404 Cyg a neutron star binary EXO2030 375 and Cygnus X 3 are modelled first provided Jan 2001 updated Feb 2002 no extended component yet SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 3 of 21 Both test cases are implemented in a 5x5 dither pattern exposure The exposure time is taken as 10 sec Background must be added such as to test the dependence of performance on signal to background ratios Validation Task Distribution Existing ISSW functions are best grouped into categories ISDAG assigned the validation task in October 2000 to Sites Individuals Data preparation tools GSFC Bonnard Teegarden Imaging tools incl response and bgd aspects generic CESR Laurant Bouchet Imaging tools survey
16. ctive observation time e Energy resolution and spectral performance e Angular resolution source separation power and location accuracy e Field of view size and performance changes e Detection efficiency e Background characteristics as far as prelaunch estimates go Test Data Two sources of test data exist 1 Raw telemetry data from existing measurements tailored for validation of some Preprocessing and Performance Analysis functions and I1 Calibrated event data from sky simulations which can be purpose made to validate specific Data Preparation and Science Analysis functions Comments Validation of tools on simulated event message data would be desirable in order to also test the data grouping and event binning functions but resources are inadequate for such a big task Event type data ideally should be simulated for a characteristic astronomical case to follow this through the different tools This involves adding time tags which correlate with pointing and their changes and ensure the consistency of event data files with pointing files and other aux data files Pierre Dubath has been generating such file groups for the simulation pipeline described earlier and will check how much he can do to import simulated events Chris Shrader together with ISDC will investigate if simulated events can be arranged such that the necessary aux data files can be generated Provision of binned event matrices and the auxilliary files associ
17. ctral Analysis Basic spectral analysis is supported through the tools mentioned above for performance analysis and inflight calibration gain analysis and background exploration Astrophysical spectral models are fitted to data by the X ray community through the XSPEC software package This package also part of the ISDC tools is enhanced XSPEC Version 12 to support the more complex inclusion of imaging and spectral responses simultaneously With this the response to each source in the field of view can be used to fold its expected spectrum into dataspace so that for the source in question astrophysical model spectra can be fitted tested in dataspace through forward folding and iteration of its parameters Before XSPEC12 is available a compromise is supported in a two step analysis In a first step of imaging analysis SPIROS is used to fit the celestial sources of the measurement and to then extract a spectrum of measured events which are attributed to the source in question In a second step these extracted source counts are then fitted to astrophysical source models with XSPEC using SPI s spectral response matrices as provied by RSGEN Note that the imaging and spatial model fitting tools can be applied in narrow energy bins This will treat the instrument response properly within each spectral bin and thus derive a proper result for each spectral bin If crosstalk among spectral bands is small as can be expected for the high spectr
18. data from surveys and open time proposals TBD The goal is to combine GPS 5x5 hex and GCDE pointings with totally 10 pointings of totally 1Msec exposure and a realistic background level CESR The performance is measured in terms of recovered source location its error and source flux amp its error shown as a function of number of pointings pointings pattern offset from instrument axis and energy Result table thd SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 14 of 21 Field of view size and performance changes The BLC exposures of a Co source at 125 m distance at aspect angles from on axis out to 10 were imaged with spiskymax Although the source locations are within expectations interference patterns between mask and detector alignments become obvious see Fig x5 Note that this demonstrates the instrument s response skymax_image_1 fits_2 Fig X5 A Co source at aspect angles between on axis and 10 degrees from top left to bottom right 0 2 4 6 8 10 characteristics if staring observations only would be used The SPI dithering observations are made to properly measure the same sky region under different aspects in order to eliminate these response feaures from the resulting image For assessment of the field of view size and the changes in imaging performance within inner and outer field of v
19. e par file typos with cryptic error messages but still producing output files see next Then several tools flood the screen with debug dump messages which do not mean anything to a general user the diversity of program log messages and debugging levels and their use is painful A most common reason for crashes is the non existence of input files or the existence of output files from a previous possibly crashed run of the tool Searching and finding such user mistakes is cumbersome and made even more difficult through uninformative error codes numerical code only by the DAL Much more user friendly DAL functionality is considered important SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 5 of 21 Specific Findings Here we list observations which apply to a specific tool we group tools according to analysis levels Test Environment Preparation and General System Use 1 User instructions were absent or scattered in places not obvious to the non regular user if the ISDC system Consult http isdcul3 unige ch Instrument spi for user instructions but this is meant for the BLC processing pipeline Consult BLC or ROT User Manual www pages follow its setup instructions create data_rep and par_files directories and the desired subdirectories Set env for login and PATH etc no idea what am doing here in detail Seems to work Copying all par files is recommended but generates a lot of unnecessary m
20. e science analysis software occurs through fitting of the amplitudes of background model tremplates prepared from above knowledge The ISSW module BGDGEN includes several analytical background model representations as well as an interface to a simulated or otherwise obtained background template also correlations to auxilliary parameters from the INTEGRAL housekeeping database e g radiation monitor countrate can be used in BGDGEN The output of BGDGEN is a background model for imaging analysis whose SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 20 of 21 amplitude should be determined however within imaging analysis to account for correlation of source signal and background in the actual measurement No special ISSW is provided for background modelling in spectral and timing domains this is addressed by spectral and timing analysis software directly ISSW tools for preparation of data for astrophysical analysis Before astrophysical analyses the measured data together with auxilliary data and responses must be collected and prepared as a data group Beyond interactive identification of the relevant data intervals and using ISDC general utilities such as OG CREATE several ISSW modules are involved here The instrument pointings on the sky are assembled by ISSW module SPIPOINT Instrument deadtimes for the time intervals in question are derived from housekeeping rates of the instrument through ISSW module
21. ess in my working directory better point to templates for general use or provide them where the help facilities are But then a Program here GENSKY does not use the current par file one needs to set setenv PFILES nobody said that before The log dump of many programms is excessive and flies by re direction to a log file may cover up the problem but how Need a system environment setup manual 2 Availability of tools From one day to the next this same setup procedure did not work any more gensky command not found search around for the gensky program do not find it nor can find a reason am stuck Some strict configuration control needs to be implemented soon 3 User Manuals of e g spidiffit spiskymax spiros are available But all are many page ps files so no edit search possible The SPIROS cookbook alone is 32 Mbytes minutes to download before one can see what it is about Attempting to invoke task help files through lt task gt h results in error messages only so cannot proceed without paper user manuals 4 General tools fv must be invoked with a strange option complication cmap 2 why this Preparation of Data for Scienctific Analyses og create Currently this uses a txt2idx preprocessor to create an index of science windows While adequate for the calibration runs where not more than a few scw s are typically combined this is adequate For real observation scenarios a more powerful
22. generates selected source spectral data or else assumes that spatial data selection and subtraction can be made to isolate the spectral signal from the source of interest For timing analyses similar considerations apply here often one isolates the source signal only through its unique signature in the timing domain and empirically defines the flat timing signature as the background from the rest of the sky and the instrument Imaging Analysis Two methods for generation of images are provided Iterative removal of sources from high to low significance ISSW module SPIROS and sky deconvolution with account for image entropy ISSW module SKYMAX SPIROS determines parameters flux significance spectrum for each identified source hence aims at point like sources primarily SKYMAX treats the sky as pixelized intensity map hence aims at diffuse emission primarily Both packages are capable of imaging the sky with point like and diffuse emission together within these constraints compromises SPIROS The package searches for a sky correlation of the instrument response with a strong point source Upon finding it the corresponding expected signal from this SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 21 of 21 strongest source is subtracted from the measured dataset and then the search is continued for the next strongest source At the end a list of identified sources exists which in the final analysis
23. iew simulated data must be used A standard 5x5 point dither pattern was used Response data were used at a grid of tbd The imaging location and flux accuracy versus source aspect angle and signal to background ratio are summarized in table tbd MPE The source flux recovery versus source aspect angle is indicated from analysis of BLC calibration runs with the same source at different aspect angles Fig X5a Recovered Source Fluxes over SPI Field of View E or gt 0 Flux cm4 2 s4 1 gt D o P N Source Aspect deg Fig X5a BLC Co source fluxes 1173 keV as recovered with SPIROS for different source aspects Source distance is 125 m SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 15 of 21 Energy resolution and spectral performance The spectrum recovered from analysis in different energy bands tests the global energy dependent analysis in imaging approaches spiros_cygius_spectra fits_1_0 GCDE 1 year 1809 keV 1 keV bir PLUX photoa an 2 s 0 001 30 100 300 1000 3000 1804 1806 17808 1810 1812 18 m energy keV SPIROS spectra from simulations of CRAB over 10 seconds Spectrum range is 40 2000 keV in 1000 LOG10 energy bins a Total detector background Jean CESR CRAB plus line features Spectrum ph cem2 sec kev 1077 107 Sensitivity is 200 4 sigma at 109 efficiency CHl
24. ld receive from MOC or generate from real data at ISDC For the completion of science validation of the SPI performance achieved through data analysis tools specific cases need to be simulated see next point Science Validation Completeness As evident from above Chapter on the SPI Science Performance Parameters many aspects have not need validated in the quantitative way needed for release of data and software to external scientists This includes most prominently the performance under more realistic signla to background ratios i e mixing background data to either BLC calibration runs or otherwise simulated data Then the performance aspects of angular resolution spectral performance and source flux recovery shall be assessed as a function of energy source aspect event type and observation patterns Remaining Activities and Planning item Action Actionee Completion Help amp Parfile Revision Oct 2002 User Support Support GUI amp Scripts for par Oct 2002 Test Data Realistic Observations ISDC w Jul 2002 MOC Test Data Sci Val Aspects ISDAG Aug 2002 Science Angular Performance ISDAG Oct 2002 Validation Validation Validation Validation Validation Patterns SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 18 of 21 5 Annex A ISSW Tool Overview We distinguish as categories of the ISSW from the user s point of view ISSW tools for data decomposition and preprocessing ISS
25. oarsely binned spihist no third party assessments yet spihisto The only significant concern we had was with the event handling logic for multiples specfically they are binned as separate events e g doubles E1 in Det 1 E2 in Det2 rather than as a photopeak event in the appropriate pseudodetector This was problematic in our response studies gensky With debug silent get lots of message dumps not at all silent GENSKY does not apparently have a manual so no help on the parameter meaning Parameters questions debug what is the difference between 1 2 which categories of output can choose from Display what is the difference between 1 2 Sources how do distinguish point sources from Gaussians Source components have a parameter spectral index and a line width what do choose for either a powerlaw or a line here respectively GENSKY ran ok but did not have the proper input file diffuse emission map and did not complain and produced an output sky image which turned out empty Or did it in the flood of dumps GENSKY does not allow for powerlaws with cut off as spectral shape need to approximate the Cyg region sources simulated by the GSFC group with powerlaws GENSKY only allows for fixed energy binwidth logarithmic binwidths are more appropriate for many steep spectrum sources The powerlaw spectra need to have proper units and corresponding normalization not always simple
26. parated by 2 and viewed by a single pointing The righthand picture shows how emulated dithering with 5 dither pointings improves imaging of these same two sources with 2 degrees separation is 36 Even when using only multiple events this degrades only to a width of 1 2 The energy dependence of this performance is relatively weak as the results at 59 keV 662 keV and 2754 keV Fig x2 indicate Rotations in viewing appear to be handled correctly by analysis software the result is unaffected Fig X3 Combining calibration exposures such that more than one source appear in the field of view the imaging results from dithering observations and the multiple source separation power was tested From the available BLC source aspects for a single source an 11 point dither pattern could be emulated Fig x3 It is evident only that dithering observations will allow the specified source separation power of lt 2 see Fig X4 A resolution power of 1 is demonstrated on these practically background free data The performance for different energies signal to background ratios and event types is tbd A test on reducing the BLC source intensity by photon sampling shows the degradation of angular performance and flux recovery with signal to background ratio see table Location Flux Flux Taking Flux Source RA_OBJ DEC _OBJ ERR_RAD uncertainty 1photonin derived signifcance 0 0 0017 0 0320 0 0038 1 6031 0 0040 1 1 603 403 5
27. pattern Derea pees Five SIGMA sources observed in core time of 4 8E 6 secs Positive reconstruction of 5 point sources near the Galactic Centre Image is a 531x531 rectangular grid of 0 5 degree B spline nodes Galactic latitude degrees 0 5 v lo 5 0 5 Detect pele Galactic longitude degrees Fig X6 Images of simulated celestial sources of the GC region sources observed by SIGMA by spiros in source left and in imaging modes right at 50 150 keV observed with a 31x31 point dither pattern with 2 separation which would be typical for the deep inner Galaxy core program observations images tbd Fig X7 Images of simulated celestial diffuse emission of the Cygnus region observed with three 5x5 point dither pattern with 2 separation which would be typical for the planned commissioning phase observations The image adopts the COMPTEL result as a model for celestial emission left The repoduced emission from spiskymax confirms that mapping of diffuse emission needs Msec exposures Here we want to use BLC data for two energies 0 5 and 5 MeV and produce images for the 1 and 2 separated cases as in Fig x4 mixing to these data typical backgrounds to simulate 10 day and 5 month exposure cases with the available 11 point dither Images would be produced for singles doubles triples and all types CESR MPE With simulated data from adopted celestial source distributions it could also be demonstrated
28. point source simulations to determine the optimal response matrix for use with spiros spectral extraction In addition GSFC will in the near future be generating background spectra to be included in the spiros KSPEC analysis The spiskymax user manual contains a very useful general instruction set on how to prepare data for testing and validations Should this be moved elsewhere into a more general user manual Imaging Sources and Surveys spiskymax Nice user manual Contains a lot of useful stuff specifically the algorithm log of an example and more examples all in one place A copy of the source catalogue is needed for spiskymax Its real use here is not obvious no directions found an attempt to edit the catalogue with fv fails file not writable SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 8 of 21 The specification of the sources for which a parameter analysis is made is irritating It is stated that the catalogue is used but why then not remove all these parameters in the par file Find out later that ONE source spec must still exist in the par file so that catalogue is used hm Once this is solved the sources are recovered and imaged as put into the simulations Outputs are fine Trying to image the GSFC simulated data fails SPISKYMAX produces empty images Using the debug flags was hopeless floods of output on the screen too fast for me spiros GSFC have run spiros in im
29. square parameter is 17164 for 17060 DOF Location 51x51 and reconstruction time of 2 mins with 1 4 GHz CPU 1077 50 100 200 500 1000 2000 Energy kev phe 21 Sep 2001 15 45 Fig Xx Spectral recoverage of simulated sources Top Left spiros reconstruction of 4 simulated sources in the Cygnus region 1 day of observation 5x5 point dither pattern Top Right spidiffit reconstructed spectrum of diffuse emission in the Galaxy viewed with the 31x31 point dither of the GCDE over 1 year of exposure Bottom SPIROS analysis of a Crab like source with line features added artificially Alternatively direct spectral fitting using the fine spectral response is performed with XSPEC This is considered the baseline approach for astrophysical studies beyond the straight data analysis presented with the imaging tools in narrow spectral bands Here we do not really have verification of ISSW as installed at ISDC only prototypes with tailored and reformatted data an assessment of real ISDC tools is tbd SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 16 of 21 Detection efficiency The detection efficiency results in the correctness of the recovered source fluxes This could be tested verified on prelaunch calibration data from BLC through comparisons of the predicted versus measured detector counts If the response matrix is correct the resultant image from deconvolution analysis quantitatively predicts the da
30. step is used as input to fit their intensities in combination through a maximum likelihood method The user may start with an expected sky which may be composed of diffuse emission and point sources also background model templates can be provided and are used in all search and fitting steps Diffuse emission and source patterns can be controlled through choices of coordinate systems spline functions SKYMAX The package iteratively modifies the inputs sky such as to improve the data fit after convolving with the instrument response using the entropy of the inputs sky as a second criterion in order to damp fitting of the noise The gradient search method is complex due to the large number of free parameters each sky pixel lterations are terminated once noise is found to dominate which detected through internal Monte Carlo simulations The user controls search method details but most importantly the entropy reference through provision of the input starting map Results are provided in the form of image and image projections Model Fitting Two ISSW modules packages are provided to determine intensity parameters of sky intensity models from SPI measurements a Maximum Likelihood fitting method and a Markov Chain Monte Carlo method source model fitting In these modules fitting methods and fit parameter constraints are controlled by the user in addition to the variety of spatial intensity model formats which can be provided used Spe
31. strated that the recovered source flux with SPIROS is retained see table 1 above in Section about Angular Performance A more refined analysis based on the known source intensities improving the accuracy of incident photon fluxes results in source detection efficiency versus energy as obtained with analysis tools as shown in Fig tbd Background characteristics as far as prelaunch estimates go tbd SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 17 of 21 Required Actions for ISSW Developers amp ISDAG Documentation and User Friendliness Stimulate ISDC to add GUI s to each major highlevel tool such as imaging analysis for SPI which provide access to functions and subfunctions help texts and result investigation methods a tool parameter database and support semi automatic assembly of dataset identifiers wherever feasible Then each developer needs to tailor the par file template to make proper use of such a GUI and guide the user into the entry of proper control parameters Also the help texts and files need to be reviewed for a more homogeneous approach now that a set of good examples exists Test Data Provision For validation of the capability of tools to handle real mission data it is required to have a test dataset which is characteristic for the real mission As a minimum the PV phase observations in the Cygnus region should be translated into the proper auxilliary files as one wou
32. ta and these can be directly compared to the measurement using Poisson statistics as statistical uncertainty measure Within statistics the match is acceptable see Figure x12 evts_det_svec fits_1_0 evts_det_soec fits_1_0 10000 6000 4000 pH t HF Ft 2000 p H H H p 0 i pie E 0 20 40 60 60 2000 4000 6000 Booo Looodg detector_id predicted counts Fig x12 The count data space counts per each of the 19 detectors tests the correctness in absolute terms of imaging deconvolution respectively the consistency of the response as used Here for the Co on axis measurement at BLC the predicted counts after imaging are compared to the actual measurement The detectors are identified in numbers for single events 0 19 doubles and triple events gt 19 left the comparison to the actual measurement is perfect right Note that the analysis includes a fit of background assumed to be constant but specific to each detector Absolute source flux recoveries for BLC data with spiros were compared with the incident fluxes as derived from analysis of a reference monitor detector with a rather well known efficiency see table Within uncertainties of both measurements this preliminary analysis indicates that absolute efficiency is properly handled Source Energy Incident Flux Recovered Flux Am 59 keV 0 681 0 725 Cs 662 keV 1 44 Degrading the incident flux of a BLC calibration run it was demon
33. that imaging tools handle larger sets of different pointings The standard 5x5 point dither pattern was tested on sources in the Virgo region while the 31x31 point dither pattern typical for the exposures of the inner Galaxy in the GCDE SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 13 of 21 part of the core program were tested on sources and extended diffuse emission in the inner Galaxy In these cases also the treatment of background was tested adopting an expected background count rate which was derived from mass model simulations Jean 2001 Figure x5 shows the Virgo region quasar images with SPI imaging tools which could be expected from the Virgo region quasars 3C273 and 3C279 Figure x6 shows the hard X ray sources in the Galactic Center region as imaged with spiros Figure x7 shows a spiskymax image of diffuse emission from the Galaxy Gatactic 6 40 20 Q 20 40 20 Q 20 40 Erdaetic i Eutactic I Fig X8 Image of simulated celestial diffuse emission of the Galaxy observed with a 31x31 point dither pattern with 2 separation which would be typical for the deep inner Galaxy core program observation This image was derived from a Gaussian shaped model left in the 100 400 keV range with the equivalent of 1 year of exposure Beyond those functional tests we simulate imaging of a sky region with a large set of separated and different observation patterns as would be typical for combining
34. the functions provided their quality and accuracy and their usability These criteria decompose into more technical aspects does each sub function execute without crashes or null results on some test case more user interface aspects can find functions and fill in their parameters and input data specs and into accuracy aspects are the quantitative results correct and consistent SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 2 of 21 The task of validation therefore is split into these three aspects e Technical validation of interfaces and basic functioning without crashes is a part of ISDC s software integration hence performed by ISDC s Instrument Specialists e User interface and easiness of use validation is somewhat subjective hence performed by SPI scientists if possible different from the software developers e Accuracy validation has many levels all to be assessed by SPI scientists The developer made unit tests of the algorithm accuracy during development using available or mocked test data which need to be complemented by broader validations within the full system using datasets prepared by other ISDC SPI tools It is the main goal to collect all this here with different issues of the document presenting more and more of the completed work early issues will list things to do We address the performance under the themes of e Energy calibration and gain correction e Dead time and effe
35. the results of performance analysis are condensed into good time intervals datafiles The ISDC package SPIGTI includes algorithms to translate SPI mode transitions into corresponding entries algorithm adaption occurs through software changes ISSW tools for near realtime analysis of bursts and transients In order to monitor incoming data in near realtime for the detection of gamma ray bursts and transients special ISSW has been tailored for the ISDC of the IBAS for realtime burst detection and QLA for daily quick look analyses to search for transient sources The IBAS ISSW features a branch for monitoring the ACS detector rates and a branch to monitor Ge camera detector rates Glitches exceeding a significance threshold above a running average are used to signify a burst The QLA software for SPI is a tailored derivative of the SPIROS imaging source search algorithm see below optimized for performance and catalogue interfacing It processes a reference catalogue of expected sources with their characteristica and outputs a list of discrepant new sources SPI ISDAG 14 05 02 issw scival_V1 doc SPI ISSW Science Validation Report Page 19 of 21 ISSW tools for instrument calibration and response determination The raw Ge detector event messages are pre processed into calibrated event messages based on gain correction factors derived through instrumental line fitting analysis For this incoming raw events are histogrammed ISSW module SP
36. the system should also cater for people who come from outside where outside means e g non X ray astronomers and or non programmers The tools themselves have often complex parameter lists whose settings are not obvious from the prompt string interactive help facilities are confined to an ascii help text file Conditional use or ignorance of program parameters makes this even more complex It seems that every task uses hidden parameters only No ISDC standard seem to apply here nor for the prompt style and value default The same is seen in the input output data spec were spiskymax prefers to provide input file specs while spiros prefers to provide data groups Generation of representative test data is a major issue obviously overlooked earlier in the project Now one must resort either to simplistic exposure patterns implemented in simulation preparation tools or be a real expert in the observation pattern implementation details of MOC and in ISDC file structures to be manually edited composed It would be desirable to have a few realistic standard cases prepared for all instruments by such experts with help of MOC cases would be the PV phase Cygnus exposure the Crab calibration exposures the GCDE core program of 1 year the Galactic plane scan Finding out about the causes of program crashes always turns out a major exercise Too little support is supplied and programs are not gracious collapsing from very simpl
Download Pdf Manuals
Related Search