Minutes of VISTA IOT meeting, May 27 2010, 09:00-12:00 Chilean time [15:00-18:00 Garching time] Minutes prepared by Steffen Mieske. Participants: Vitacura: Thomas Szeifert, Steffen Mieske, Alain Smette Paranal: Valentin Ivanov Garching: Jim Emerson, Michael Hilker, Marina Rejkuba, Wolfgang Hummel, Lander de Bilbao On the phone from the UK: Jim Lewis & Mike Irwin [Resulting action items are listed at the end of each discussion item] 1. VISTA status and planned activities (TSz, VDI) (intervention 24.04-5.05.2010). Thomas Szeifert summarises the current VISTA/VIRCAM status, which is not too good, given many problems with data quality. - The shimming was changed during the intervention and now there is an IQ gradient across the FOV resulting in 4 of the 16 detectors being distorted / out of focus. - The tracking performance is reduced in altitude direction probably caused by a faulty gyro which protects against over velocities of the drive. This results in significant ellipticities. - The preset sequence is taking the images of the detect guide star too quickly and starts the guiding too quickly. This results in elongated images especially for short exposures. Assistance is needed for resolving the problems with shimming and altitude axis (gyro). We are waiting for feedback from Vertex reg. gyro. Thomas Szeifert notes that it may be considered to stop taking data, and that this will be discussed with Paranal engineering in the first week of June, when the respective contact persons are on the mountain. The possibility to stop taking data is generally discussed. Valentin Ivanov notes that this would increase the pressure especially on software, who had recently been overwhelmed with other problems on Paranal. He will approach Pedro Baksai regarding the software related VISTA/VIRCAM problems. Marina Rejkuba notes that it is difficult to judge which survey is affected most. The checks done up to now were performed on a small sample. Having said that, she would not oppose stopping the operations. [Note added in proof: 3 days after the meeting, Valentin Ivanov inspected the opslog data derived by the Paranal pipeline for detector 11 over a long time scale. The conclusion clearly is that only the VVV data are affected by high ellipticity (which 30-50% of the images). This is valid for _no_ deblending performed by the pipeline. Jim Emerson notes that the statistics may look more favourable once deblending is switched on. This was done on May 31.]. Wolfgang Hummel also notes that 30-40% of the data cannot be processed because of the CRVAL problem in the extension headers. Jim Lewis notes that this is not a fundamental problem regarding data quality, but is something that is fixed by the CASU during the processing. Thomas Szeifert asks Valentin Ivanov to check with Gerd Hudepohl whether there is an update on the Gyro issue. He is waiting for advice from Will Sutherland regarding the shimming issue. A horizontal shift of the camera may cause the focusing problem. Jim Emerson notes that he will talk to Will Sutherland but that details will be worked out by ZEMACS. Thomas Szeifert notes that putting a pause in the preset software may help to remedy the problem with guiding before tracking. He also notes that the problem has existed for some time and that he had before blamed M2 for it, although now it turns out that it is the altitude axis. Jim Emerson notes that regarding the guiding problem he will inquire with Steven Beard regarding its history. Marina Rejkuba notes that it may be necessary to revise the classification rules for OBs taking into account ellipticities. Thomas Szeifert notes that the pipeline is not too reliable regarding ellipticities in crowded fields since no deblending is done. It is agreed that deblending shall be switched on asap [done on May 31] and that the ellipticity results should be reanalised afterwards. Resulting action items: AI 2010-12 TSz: send to JEm copy of PPRS regarding guiding/tracking problem. AI 2010-13 TSz: send e-mail about result of discussion with head of Paranal engineering regarding the shimming/gyro problem AI 2010-14 TSz: if necessary, have e-mail discussion on whether stopping VISTA operations (during first week of June) AI 2010-15 VIv: check with Vista Managaer on update on Gyro from Vertex AI 2010-16 VIv: contact Paranal software to doublecheck on progress of CRVAL problem AI 2010-17 VIv: notify operators of the ellipticity problems such as to favour longer exposures for the time being AI 2010-18 JEm: check with Steven Beard regarding the history of the guiding/tracking problem AI 2010-19 JEm: discuss shimming problem with Will Sutherland AI 2010-20 LBi: enable image deblending of the paranal pipeline [done on May 31st] AI 2010-21 WHu: check image ellipticity on the next data delivery 2. Persistance (WHu, TSz) Thomas Szeifert notes that resetting does not change persistence. Visual inspection of test images gives the following impression: the count decay should be same in time for a sequence of long darks and for 100 times resetting. There is probably no way to get rid of this. It needs to be checked whether user documentation is taking into account appropriately this effect. Thomas Szeifert and Jim Emerson note that users should be notified about which template settings will result in persistence that is not removed in stacking. The users may have to accept slightly larger overhead to get around this (holding especially for VVV) Mike Irwin / Jim Lewis: Whether or not persistence is visible in the reduced data depends on the nesting algorithm (e.g. it is there for the NB SV data for N253). When using the FPJ sequence, then the persistence images get rejected during stacking process. So for most users persistence should not be a problem. Mike Irwin notes that VISTA data will be reprocessed with the pipeline version 1.0 with improved persistance rejection. Marina Rejkuba notes that there was a specific reason for using the particular nesting sequence in SV data. It works as long as there are enough random jitters (for J and Z). It does not work well for NB because there was no jittering. She also notes that there are currently no NB OBs in the queue. Mike Irwin notes that Ultravista data are fine. The persistence problem seems most peculiar to GTO time and SV data. Valentin adds that the VVV stack together variability frames of the same dither pattern which may also cause problems. Mike Irwin adds that one will discover small remnants if there is no random jitter pattern used. AI 2010-22 USD: contact affected PIs about persistence consequences. 3. VISTA Selected PPRS Summary (VDI) and operational issues 033950 Instrumentation Assigned detector 16 is badly unstable --> TO BE CLOSED. Nothing to be done about it. 035169 Software Assigned VIRCAM OBs with CRVAL1,2=0 in extension header --> Ongoing. VIv: contact SW group leader to ask for progress 033084 Electronic New M2 still moving after the jitter offsets --> Ongoing. Was discussed before under discussion item 1. AI 2010-23 VIv: close PPRS 033950, add note in documentation AI 2010-16 VIv: get back to SW group leader re PPRS 035169 [see also Discussion item 1] 4. Progress of the surveys (MRe) Valentin Ivanov states that his survey execution simulations show a 85-92% completion rate for the current semester. However, neither the intervention nor a possible stop of observations was included into the simulations. According to the simulations there will be some idle time in July/August, but this will likely not be an issue given the current situation. Marina Rejkuba notes that before going for carry-under OBs, we need a comparison between the simulations and the actual progress. It is agreed that Marina sends her actual progress data to Valentin for him to compare with the simulation data. Run priorities: Thomas Szeifert suggests separating the Ultravista P86 OBs into two programs/runs with 0.8" and 1.0" seeing respectively. Marina Rejkuba summarises that Ultravista requested 360 hours per year in the Survey Management Plan. Scheduled were 100 hours in the dry runs and 160 hours in the first semester --> 260 hours in the first year. She summarises her communication to the PIs, after they had expressed concerns that their survey was not completed enough. She notes that they will find it strange if in the end they get less data than with their original strict 0.8" seeing constraints. The OT ranking priorities are discussed. One VHS field collides directly with ultravista in those priorities. It is suggested that ultravista should get higher priorities since there are more VHS fields. Valentin Ivanov and Marina Rejkuba summarise the underlying technical problem, namely the assumption that the most important programs would also have the most stringent constraints. This led to confusion when Ultravista degraded their constraints from 0.8" to 1.0". Thomas Szeifert strongly states (clapping his shoe on the table) that these relative priorities should be fixed in phase 1, not phase 2. A fudge factor is needed for re-scaling the ranking of the runs according to their individual progress. This factor should be implemented into a database (feedback loop: survey progress --> OPO changes fudge factors --> OT ranking automatically updated) and read in as an additional parameter into the OT ranking. It should hence not be necessary for VISTA operations team to 'manually' change anything in the OT ranking algorithm. AI 2010-24 MRE: send survey progress data to Valentin Ivanov [note added in proof: DONE - MRE sent data via e-mail to VIV on 2010-06-01] AI 2010-25 ViV/GHu: compare actual survey progress with the simulations AI 2010-26 Tsz: prepare proposal how to implement fudge factors into the OPO/ESO database for re-scaling OT ranking priorities of the various surveys. [note added in proof: ONGOING - TSZ has sent a recommendation memo to MRE, who discussed it with Gaitee Hussain (OPO). OPO has to decide how to implement this also in view of upcoming complications on scheduling Survey + GO observations.] 5. Photometric zeropoints from science frames (several related tickets; WHu, VDI) Valentin Ivanov found that many of the ZP distributions in a given chip are asymmetric around their mean (using the Paranal pipeline). Mike Irwin states that he did similar tests with the CASU pipeline which look reasonably well in terms of symmetry. He notes that ZPs in JHK are consistent with the expected rms, the Y-band is also close to OK, and that only the z distribution is broader than expected, albeit symmetric. Regarding the long-term trending of ZPs, Thomas Szeifert finds now a 0.10-0.15 mag fainter instrument response than directly after the coating. However, there is no new coating date foreseen in the medium term future. Mike Irwin states that those differences are leveling now to a constant value, and that K-band ZPs vary less than J. He finds that the effect is 5% in K and 10% in J. Wolfgang Hummel reports on the significant variation in ZPs he had found between subsequent paws of one tile (fixed detector). Mike Irwin notes that he will look into this in the next few weeks. Regarding the trend of ZP with crowding, Mike Irwin states that deblending and extinction correction help removing this trend. Steffen Mieske notes it will be important to check how the Paranal ZPs evolve after enabling deblending. AI 2010-27 TSz: check ZP evolution for all filters. AI 2010-28 WHu: send to Valentin Ivanov plots on the ZP checks. 6. QC monitoring scoring of opslogs values (finally operational!) (WHu) Closed, this is OK. Thanks to Wolfgang for implementing this! 7. Re-processing of VISTA data by CASU (JLe) Mike Irwin / Jim Lewis: re-processing of all the data with bug-fixed pipeline is imminent. 8. Current status of the header keywords for CASU pipeline reduced data (MAr): "I would like to enquire whether Jim Lewis did implement the suggested changes requested by Adam to the headers keywords, to ensure the uniqueness of the "extname" extensions in the *_st.fits files." Adam's report is enclosed bellow. Jim Lewis / Mike Irwin: This is fixed in v1.0 of pipeline. 9. Feedback from the PIs (MRe) Marina Rejkuba summarises the feedback she got from VMC, Ultravista, VVV PIs. Michael Hilker has had no feedback from VHS and VIKING PIs. Some smaller concerns come from VMC and VVV PIs which were not discussed at length (problems only for small subsets of data). A possibly more serious concern came from Ultravista PI regarding the too small random jitters, but they are waiting for a larger batch of images to make more definite conclusions. Regarding the ultravista concern, Valentin Ivanov inquires whether the seq file for jitter creation checks for too small jitters next to each other. Jim Emerson replies that this is not the case. The algorithm was taken from HAWK-I. AI 2010-29 VIv: check exactly what the seq file does for VIRCAM jitter 10. Action Items review (enclosed below) 11. Others Marina Rejkuba notifies that the tentative data for the webletters is 23rd of June. Jim Emerson briefly summarises the changes to the new SADT version. Regarding SADT output, there is only one change, namely in the jitter. The NB980 filters are briefly discussed. Marina Rejkuba has had an initial discussion with Thomas Bierwirth and Dario Dorigo about including a solar elevation criterion to the OT ranking. She has not heard back from them since. AI 2010-30 MRe: Check the EVM based on changes done to SADT. ETRM is looking fine. She will try to test the ETC wrt real data. 12. Date of the next meeting (end of June, 2010) June 29, exact time to be verified since Jim Emerson will be in California. ************************************************************ Review of action items from previous IOT meeting (2010-04-23): Progress report on previous action items: * AI 2009-01: Monitor execution times & survey completion status is ---------- being followed up by MAr, MRe. 2009-07-29: there is a first prototype 2009-10-06: no news (MRe) 2009-12-01: no news 2010-03-23: the progress was reported earlier (see above; MRe) let's have another review on the next IOT when we will have more OBs executed (MAr) 2010-04-23: no news X-Original-STATUS: ONGOING [update 2010-05-27] CLOSED, is a permanent issue * AI 2009-11: TSz should find out if the constraints sets could be carried with the OBs, to the OS and the fits header of the VIRCAM files. Otherwise it is not easy to propagate these data by interfaces. 2009-07-29: TSz iterating with R. Schmutzer, but DFI (T. Bierwirth) will also need to be involved. For surveys, given the small range of programs it is expected that there are not too many different OBs, and this might need to be followed in a different/manual way in the beginning - this could imply a major change for the tools that cannot be done in the last moment. 2010-03-23: the keywords have been implemented except for the moon; there is a ticket with details; some testing done by VDI, no errors found; keep ongoing, try to complete during my next shift end of Mar - early Apr. 2010-04-23: VDI reports one formula is missing X-Original-STATUS: ONGOING [update 2010-05-27] Corresponding PPRS is still pending. * AI 2009-14: GHu to check that the seeing distribution function in the scheduler for VIRCAM takes into account the finite (0.51") nominal resolution of the instrument. 2010-03-23: GHu reported that it is not clear if TaToo will be used for scheduling time on VISTA; decided to keep this AI open until it becomes clear how VISTA time will be scheduled. 2010-04-23 no news X-Original-STATUS: ONGOING [update 2010-05-27] Ongoing. Thomas Szeifert states that for the scheduling the seeing distribution is _not_ considered in the requested band. Marina Rejkuba notes that there may be some chance that this is updated by ESO. * AI 2010-01: TSz the calibration database should be updated with new flats and master calibrations and how to do it regularly, as no procedure is currently in place. 2010-01-26: WHu should give Thomas Bierwirth a USB disk with the updated calibrations before Feb7, 2010. Then on try to define a procedure for a regular update of reference frames on Paranal 2010-03-23: done in Feb; the frequency of updates should be three months, the next one will be in May; SMi will talk to DA's to find out how to clean the database, i.e. removing the old calibrations with a script? Once this is done, we close this AI. 2010-04-23: given the intervention, it is not necessary to update the calibDB now with pre-intervention calibrations. X-Original-STATUS: ONGOING [update 2010-05-27] Closed. New DB prepared by WHu (May 21). * AI 2010-02: TSz: how many standard stars fields should be taken? Effort in February to take up to 3 fields. Request to implement a change of jitter pattern by JLe, from 4 to 5 jitter. Try to keep calibrations plan withing 1 hrs total execution time. 2010-01-26: JLe suggests to discuss issues with PIs, and provide feedbacks to TSz, especially in the ZY band calibrations in particular. 2010-03-23: MRe had a discussion with P. Lucas (VVV co-PI) about the frequency of the standards, the feedback is pending, we will keep this open until we hear from him. VDI points that we will overcalibrate at the beginning but at some point we won't need to bother anymore with standards at all because we will have the entire sky calibrated in the NIR (unless one needs better accuracy or in special cases of high extinction regions). A short discussion about feeding this back to the PIs - at the moment it is premature (as well as the information about the expected extra year necessary to complete the surveys) because we are still unsure ourselves. 2010-04-23: no news (MRe not attending) X-Original-STATUS: ONGOING [update 2010-05-27] Closed. Situation cannot be improved with the current set of standards. Corresponding discussion: Marina Rejkuba summarises comments from the user committee. The SMP of VVV asked them to take into account additional time for standards since they do heavily extincted fields. Then ESO decided that users are not charged standard star OBs. The system is calibrating itself, so it should be fine in the end. Thomas Szeifert notes that only the VVV asked for additional standards, doing UKIRT in addition to TSF. They do one UKIRT per night, on one chip only. This needs extrapolation to other chips. He also notes that a 'Landolt-type' project with several FTE would be needed to provide a consistent set of standards with large colour range in the 'non-standard' filters. Mike Irwin notes that it is up to the VVV team to make suggestions on how to optimize their photometric calibration. They can do precise internal, relative, VISTA calibration, but no absolute calibration. * * New Action Items from 2010-03-23: AI 2010-03: completion rate of the surveys to be compared with the new simulation of scheduling which should now include real submitted OBs. This comparison should be used to review the ranking and decide if it is necessary to revise it for the future. MRe to provide completion rate until May 1. GHu+VDI to run a new simulation with realistic OB database. Deadline for this AI is May IOT. 2010-04-23: no news [update 2010-05-27] Ongoing, to be redone including technical time loss. AI 2010-04: TSz to modify the AO priority/validity 2010-04-23: no news (TSz not attending) [update 2010-05-27] Tolerance changed. Closed. AI 2010-05: JLe to incorporate the EXTNAME in the CASU products 2010-04-23: no news (JLe not attending) [update 2010-05-27] Done. closed. AI 2010-06: TSz+VDI to improve the TIO training, to prepare an operational manual 2010-04-23: done X-Original-STATUS: ONGOING [update 2010-05-27] Operations wiki has been updated. More training needed. Ongoing. AI 2010-07: JEm to fix the SADT - 2010-04-23: JEm will fix the most critical items on the bug list, but does not guarantee to fix items that go beyond the original requirements. Some critical items are to be discussed, like the control of parameter values that might be manually edited by the user. The PA bug was identified and will be fixed. The new catalog issue on UCAC-3 isn't solved yet. The tool will be delivered by the end of next week. Ancillary material (documentation) will be distributed. [update 2010-05-27] Ongoing. New corrected version of SADT released, testing by ESO staff until June 8. Summary of corresponding discussion: Jim Emerson notes that tile2 is still not treated properly. It is commonly decided to remove this from the IP for P86. SADT also now reads UCAC3. He states that the remaining issue with proper motion are an issue of athe catalogue, not SADT. He furthermore suggests to spend a night of engineering time to check well what are the correct suitable magnitude ranges for Technical / Guiding chip. Keeping the same allowed magnitude range at a different catalogue (UCAC3) may not necessarily help. Thomas Szeifert adds that there can be very high magnitude uncertainties in the catalogues. For Ultravista the problem is that possible guide stars are generally on the brightness limit. Marina Rejkuba inquires whether the UCAC3 will help Ultravista. Jim Emerson replies that there are less stars in UCAC3 than in GSC. Jim Emerson also asks to send comments/objections to the modifications done to SADT, which are summarised in his recent e-mail. Michael Hilker ask Valentin Ivanov to do SADT tests on the sky during the next week or so. He will try to do this, together with the IP test. Michael Hilker asks for comments regarding SADT until June 8. AI 2010-08: VDI to take care of the new IP by May 2010-04-23: will be done X-Original-STATUS: ONGOING [update 2010-05-27] Ongoing. IP prepared, test is still pending in combination with new SADT. ** New Action Items from meeting 2010-04-23: AI 2010-09: photometric zeropoints as calculated by the pipeline. [update 2010-05-27] Ongoing, see corresponding discussion this meeting AI 2010-10 WHu: It should be investigated, if the quartile of the ellipticity can be used as a threshold for OB rejection. [update 2010-05-27] Ongoing, to be investigated after new de-blending option AI 2010-11 LdB: use a new default value for the de-blending option of the pipeline [update 2010-05-27] Done [note added in proof: done on May 31].