CXC response to CUC Recommendations following 2009 Apr CUC Meeting > 1. Calibration. As mentioned above, while the agreement in cluster > temperatures between Chandra, and XMM and Suzaku appears to have > improved with the latest updates, the 0.5 - 2 keV Chandra flux now > appears to be in disagreement not just with XMM and Suzaku, but many > other x-ray astronomy missions as well. While the CUC do not consider > that agreement with other missions should be the ultimate goal of > calibration, and endorse the approach of the Chandra calibration team > of trying to understand the instrumentation in as much detail as > possible, we remain concerned at the continuing problems surrounding > the Chandra HRMA calibration. We cannot really find fault with the > technical approach, but wonder if better communication with the user > community could be facilitated. > > RECOMMENDATIONS > > 1.1 Giving the changes in fit parameters for e.g. cluster source and > AGN with power laws (extreme thermal and nonthermal sources) for each > change in the CALDB. This would allow users a clearer idea of by how > much their fit parameters could be in error due to the calibration > uncertainties. When cross-calibrating between different X-ray telescopes, it is also important to cross-calibrate between each detector and grating combination, since all telescopes have some internal cross-calibration uncertainties. For example, the recent paper by Mateos, S., Saxton, R. Read, A. \& Sembay, S. 2009 (A&A, 496, 879) notes that the fluxes derived by the XMM-Newton MOS cameras are 7-9% higher than the fluxes derived by the PN camera below 4.5 keV. At higher energies, the flux discrepancy between the MOS and PN is 10-13%. The most recent cross-calibration results between Chandra and XMM-Newton were presented at the 2009 International Astronomical Consortium for High Energy Calibration (IACHEC) meeting by Smith and Marshall (see www.iachec.org/meetings/2009/index.html) They presented the cross-calibration results for a sample of AGN observed by both the Chandra gratings and XMM-Newton. This comparison shows that there are some energies bands where there is good agreement between the Chandra gratings and the XMM-Newton detectors and gratings and some energy bands where there are discrepancies up to 10%. The most recent cross-calibration effort between Chandra and other X-ray telescopes (XMM-Newton, Suzaku and Swift) using ACIS imaging data was presented at the 2009 IACHEC meeting by Plucinsky et al. 2009 (SPIE proceedings) based on observations of the supernova remnant, E0102-7219. This comparison shows that the fluxes in the OVII and OVIII lines (0.5-0.7 keV) derived from the ACIS data are within a few percent of those derived from the RGS and MOS data. For the NeX (1.0 keV) lines, the ACIS-derived flux is 10% higher than those derived from the MOS detectors and the Suzaku XIS0, 20% higher than the flux derived from the PN, but in good agreement with the fluxes derived from the Suzaku XIS1 and Swift XRT. Thus, there is not a systematic offset in measured fluxes between Chandra and all other X-ray telescopes in the 0.5-2.0 keV band. Rather, there is approximately a 20% scatter in the fluxes derived in the 0.5-1.0 keV energy band with the present fleet of X-ray telescopes. The Chandra calibration team has not yet performed a more extensive cross-calibration effort between Chandra and other X-ray telescopes in the 0.5-2.0 keV band due to the upcoming improvement in the ACIS contamination model. Once the new ACIS contamination model is released, the Chandra calibration team will perform a more extensive cross calibration with other X-ray telescopes using clusters of galaxies, AGN and supernova remnants. > 1.2 A significant update to the instruments and calibration section on > the Chandra website is needed. The initial table describing all > calibration topics and their current status is useful, but there is no > way for a user to know which topics are under active investigation; > some bolding or use of color would help. At the next level down, > however, we note that old archived powerpoint presentations are not > adequate. We suggest written memos, possibly short, that describe the > issues being worked by the calibration team. These can be then > categorized as "Open Issues", "Best Working Guess" and "Obsolete > Items" (or similar). Short reports on the HRMA A_eff (re)calibration > and ACIS contamination represent especially important and timely items > to include in the revamped Cal section of the (newly reworked) CXC > website. > > 1.3 Working with the CIAO team to annotate the CIAO analysis threads > with warnings regarding the calibration at the appropriate > places. Many users may just skim the real documentation (we know they > should read it properly, but this is human nature!) and warnings in > the threads would force them to take notice. The calibration team is presently refurbishing the Chandra calibration web pages, but, as the CUC suggests we also need to add some new sections regarding hot topics and how the calibration uncertainties can affect an observer's scientific conclusions. We will present some prototypes for these new sections at the next CUC meeting. > 1.4 The HRMA discrepancies were found and certainly addressed as > an on-going effort of the IACHEC group that looks into the > cross-calibration of several X-ray missions. We strongly recommend > that CXC continues its involvement in this consortium. The Chandra calibration team will continue to support the IACHEC. > 2. Software. As with the calibration, we have no real complaints about > the technical quality of the work of this group, but similarly do have > some concerns about their relationship to the user community. We now > have two versions of CIAO, 3.4 and 4.1 that are currently widely used, > and suspect that despite the 731 downloads of CIAO 4.1, the vast > majority of users are still working with CIAO 3.4, largely due to the > overheads associated with learning the new interface. We make the > following recommendations: > > RECOMMENDATIONS > > 2.1 Survey current CIAO users to determine which version they are > using and why, along with what they would like to see included in > the next versions. A selected sample of recent authors of Chandra > data based papers would work, or some other criteria to avoid self > selection bias -- follow up may be needed to get a complete, or > near-complete survey. As one example, the CUC believes that the > software industry is going to go with Python and recommend that the > Chandra software do the same, but this would be a good question for > the survey. SDS developed a user survey and released it to the CIAO community. The survey was advertised on the CIAO web pages, the chandra-users, email list, the electronic newsletter, and on internal email lists. We have 131 responses so far (about 90 from outside CXC) and responses are still coming in; we will analyze them at the end of July. Decisions on scripting languages and other CIAO development goals will be made once the survey results have been assimilated. > 2.2 Make more example files available for the new interface. The > service available to translate users' scripts is valuable, but the > Chandra SDS group should understand that many people will be nervous > about revealing their programming techniques (or lack thereof) to > software professionals. [Some authors of this report would certainly be > in this category!] In addition, even if the translation were perfect, > the user would still have to explain the purpose of the code to the > team, then test the results, and finally understand how it works -- a > significant effort, even if less than rewriting it themselves. Making > (many!) more example scripts available would allow users to make their > own substitutions or translations as needed. Additional example files have been added to the 'gallery' pages. http://cxc.harvard.edu/chips/gallery/thumbnails.py.html These include the python scripts used to make each plot. We are investigating a similar approach for Sherpa and will report on this at the next CUC. In the meantime we have added several new threads to http://cxc.harvard.edu/sherpa/threads/all.html > 2.3 The SDS need to market their software more effectively to the > Chandra community. Something like an EPO effort (although of course > this is not really public outreach, but outreach to professional > astronomers) to motivate users to upgrade to CIAO 4.1 and Sherpa seems > to be required. We are really suggesting that more effort be put into > making better use of the existing functionality of the software, > rather than developing more. We will develop a plan to do this based on the survey responses; we will advertise useful features of CIAO in the newsletter and other venues. Kenny Glotfelty of the DS team has developed an excellent new CIAO demo package which we presented at the summer AAS and we will look at making this widely available. We are also working on additional scripts to simplify aspects of the analysis as well as improving existing scripts. We will continue to offer the CIAO workshops which have proven very useful. > 2.4 It would also be good to hear (for example at the next CUC > meeting) about CXC medium and long term plans for CIAO: which > platforms will be maintained for how long, what is the planning for > the time after the operational phase (i.e. post-existence-of-CXC, > will the number of builds have to be decreased, are there any plans to > go for virtual machines, grid solutions, etc)? Our medium-term platform support plan is to reduce the number of builds to one each for Linux, Solaris, and Mac; the Data Systems team are experimenting with CentOS as a way to maintain a Fedora-compatible build that doesn't require frequent updates. Support for specific OS releases is always a short-term decision, and we take care to keep abreast of what the community is using. For the longer term (post-mission) we are identifying a 'minimal CIAO' which would probably include the process_events tools, the response-building tools and their supporting libraries. We conclude that post-mission, the latest archive reprocessing will be deemed definitive, so early pipeline tools will not be needed, and software advances will mean that the late-stage analysis tools, non-Chandra-specific, will either be superseded or new versions will be funded via other missions. But there will still be a need for archival users to filter, tweak, and otherwise reprocess the evt2 files, and to extract products (spectra, light curves,etc) and generate the associated response files. The DS team has already made great progress in modularizing CIAO so that subset releases can be supported, so much of the infrastructure is already in place for such a minimal CIAO. > 2.5 Put mean/median helpdesk response times on the Chandra website so > that people know they are fast. Done.. http://cxc.harvard.edu/helpdesk/ now says: In most cases you will receive an initial response to your ticket within two working hours of the time of submission. > 3. Catalogs: As mentioned before, the presentation of graphical > interfaces for accessing source catalogs generated much interest and > discussion among the CUC. So far, these interfaces all seem to be > isolated efforts (the Chandra Source Catalog, TGCat, the JUDO and UDON > interfaces developed for Suzaku) so: > > RECOMMENDATIONS > > 3.1 At a minimum we recommend that the various catalogs and interfaces > have links to each other. In particular a link to SIMBAD would be > desirable. Since most astronomers use this to search for sources, it > would be a shame if Chandra sources did not show up in a SIMBAD > search. It is also particularly important for a user to be informed > that just because a position returns no 'hits' in the (evolving) > Chandra source catalog does NOT mean that that position has not yet > been observed by Chandra. In the longer term the CUC would be very > excited to see a uniform graphical interface that allows access to all > of the data of interest of a particular source from the various > missions. Website links - As a matter of convenience, the CSC team can add links to various other catalogs and interfaces to the CSC public website. The TGCat main page (tgcat.mit.edu) has prominent links to: CSC, X-Atlas, BiRD, HotGAS, and MAST, which we considered to be the most relevant links for Chandra data and high resolutions spectroscopy. These are links to the catalog main page, not to TGCat sources within the given catalog. Additional access - We are presently working with several groups to provide access to (a subset of) CSC data via interfaces other than CSCview, including SIMBAD and VIZIER, HEASARC Browse, and standard VO interfaces. These interfaces will likely be limited to accessing catalog Master Sources Table data, since they generally cannot support the many-to-many linkages between the table of distinct sky sources and the table of source detections from individual observations, or access to the FITS data products. We expect most of these interfaces to be available before 2010. In addition to CSCview, CSC data can currently be accessed directly from ds9, and the CSC has a registered IVOA standard Simple Cone Search service, which enables VO-aware applications such as DataScope, TOPCAT, WWT, ..., to retrieve tabulated CSC source data. A CSC-SDSS cross-match should be available publicly very soon. The cross-match will be released simultaneously by the CSC and SDSS groups, and hosted at both CXC and JHU. The release date depends on the completion of a memorandum of understanding between the two groups, but should occur sometime in summer 2009. The cross-match applies to a specific release of the CSC and a SDSS specific data release, and we will aim to provide updates whenever these change. The TGCat search results tables contain a link to SIMBAD for each extraction, as well as to the Chandra Obscat ("Links" column). During the V&V process, uniform names are assigned, since the observer-specified target name may vary for the same object. We generally use the primary SIMBAD name. The TGCat "Object Type Search" uses SIMBAD object classes and lists the number of sources in the catalog of each type. Sky coverage - We are in the process of deploying a beta version of a Google Earth interface that will allow users to visualize directly exactly what regions of the sky are covered by observations included in the CSC. Extending this to also include the fields of view of observations NOT included in the CSC is reasonably simple, and we are working to add this in the short term. However, the most reliable way for users to determine whether a particular location has been observed with Chandra is via a footprint service. The latter is beyond the scope of the CSC, but a CXC project is currently underway to make such an interface available in the next few months. As an follow-up enhancement to the latter, the CSC team also plans to use a similar mechanism to provide the local catalog limiting sensitivity in each energy band. Web-page access to a prototype (lower spatial resolution) limiting sensitivity service is almost ready for release. Future uniform interface - We agree that a uniform graphical interface that allows access to data from the various sources (from all wavebands) would be a valuable asset. Several members of the UK AstroGrid group at the University of Leicester are particularly enthused with the CSCview interface. At their request, we are currently exploring a co-operative effort to use CSCview as a stepping stone to create a standard generic query builder that could be used for generic queries and data mining for catalogs that support VO standard interfaces. > 3.2 The user interface to the catalog should be reconsidered, perhaps > with advice from the ds9 developers -- who themselves have a goal of > immediately-usable software. The current interface, tool tips and > help, and error messages need quite a bit of attention in the very > near term. Most of the CUC members who have tried the interface have > found things confusing or gotten unexpected results. It is important > to address the new-user experience now, since the first impression may > drive away many potential users. The HEASARC Browse interface will > provide a base level that most users are familiar with, so the goal > for the CSC should be that an astronomer be able to immediately use > the Chandra catalog (without reading any documentation) to get more > useful information than will ever be available in Browse. New user experience - The CUC comments were written before the latest release of CSCview, which includes several improvements that address many of the issues that were impacting new users of the interface. The team has also recently embarked on an analysis of how new users interact with CSCview, by watching a range of in-house scientists (from students to senior astrophysicists with a range of computer skills and experience) use the application for the first time. With the recent CSCview update, all of the test subjects were able to use the interface to perform simple queries, although several enhancements were identified and will be folded into subsequent CSCview releases. A remaining area of concern is better support for access to metadata describing catalog columns (descriptions, data types, etc.). The public website does include this information, but next release of CSCview will include further enhancements to simplify access to the relevant data. > 4. Website. Finally, we just add a 'kudos' on the new website > design. We encourage the CXC to go public with the new design ASAP, > and to view its release as an opportunity to make major facelifts to > individual team pages (the Calibration portion being a problem > area, as noted above). The new front CXC webpage will be posted before mid-August, with all links checked to be live, and obsolete pages removed or archived. We are working with all the CXC teams to review and update their pages. Key updates from this longer-term effort will be available for review before the next CUC meeting.