Report on calibration as of 2003-01-29 -------------------------------------- The calibration of the NPOI 6-way data continuous to be no better than 10 - 40%. Phase calibration suffers from incomplete bias correction. The standard reduction includes the following: V2 bias correction in CONSTRICTOR using the amplitude at a specific user-defined k frequency (k=4 for 3-way, k=8 or 10 for 6-way), or a fit to the amplitudes at up to three different k frequencies (e.g. 1, 7, and 10), all done directly with the coherent scan data. Closure phase bias correction for those triangles with all baselines from the same detector. Based on Poisson assumption. Mixed case has smaller bias and is not corrected. Additional OYSTER V2 bias corrections using incoherent scans, only applied to the single k frequency compensated scans. The scans with fitted bias do not need any additional bias compensation. V3 bias correction in OYSTER using incoherent scans. Detection of sub-array configuration by checking the total NAT counts for each station and scan. Identification of stations in search mode by 2 ms lag median absolute delay change encoded in FDL delay error bar. This is important for those scans where the "way" parameter set by the observer is less than the number of stations used. Calibration versus time. Standard reduction does not include: Bias correction of mixed triples, probably negligible, but analytically unknown so far.. Non-poisson bias correction of closure phases, probably important for faint stars.. Photometric calibration, which is only available averaged over the entire night. Refitting of variable position fringe peaks, i.e. non-integer k frequencies. This problem is better addressed in the control system; in post-processing very expensive to treat. Pair-wise bias correction for non-fitted scans. Issues which need to be addressed, beside the one mentioned above: Priority 1 (highest): --------------------- Investigation of archival data from 2002, which remains largely unreduced at the required level. Consistency checks with respect to throughput, system visibility, background, etc. Systematic errors in the back ground data, related to detector non-linearities, sky contamination near bright stars. Perform tests using new delay between scan and background measurement. Photometric inconsistency between photometric scans and full array scans. Priority 2: ----------- Fractional corruption of data by micro-locks. Editing of raw FDL delay excursions and spikes. Throughput differences between stations and their dependence on wavelength. Simulator for after-pulsing and investigation of proper bias correction for triple products. Priority 3: ----------- Investigation of different seeing indicators. New ones are available through OYSTER, but have not been tested extensively enough. Best scheduling compromises between number of coherent and incoherent scans. This depends on the actual benefits currently realized for the calibration. Assessment of differences between data requested and data received due to observational problems. "Starwheel" scheduling software. What has been done on calibration during second half of 2002: Detector non-linearity tests and their reduction (with DM) Fringe peak spectra discovered as a tool to check data quality. More routinely computed and posted on the internal web later in 2002. Bias correction added for triple amplitudes (V3) Fits to unused fringe frequencies fo bias computation. Little empirical data yet due to delays in data analysis. General problem: tools are provided, but not much time left for me and due to lack of others using OYSTER to gathering empirical experience for data analysis. What did I do in the first half of 2002, before calibration was investigated more closely: Modifications of data analysis for actual 6-way observations. At the same time, making some updates to manuals and web exercises for OYSTER to enable others to follow. Eta Virginis data reduction and paper. During this analysis calibration was recognized as a problem, but I was able to deal with it using baseline fudge factors. This method had been used before and I deemed it reliable, more or less. Astrometry analysis. No time for much else. How can we make progress now: The above issued need to be addressed, whereby the challenge is to have good bookkeeping for comparative analysis. For example, compare different methods of bias correction. Compare quality indicators for different nights and see effect on data. All results should be documented on the NPOI internal page. Since we now have, in addition to our astrometry task, the imaging project, an additional full time astronomer is needed. This person, to be stationed in Flagstaff, would agree to learn and become very familiar with the standard NPOI data reduction software (e.g. OYSTER). He /she would be expected to write code for OYSTER, update the manual on new procedures, provide feedback, and take on other routine service tasks such as daily reductions, and scheduling. -- Dr. Christian A. Hummel NPOI Division, Flagstaff Station US Naval Observatory c/o 3450 Massachusetts Avenue NW Washington DC 20392 Tel.: 202-762 0314 FAX: 202-762 1514 cah@usno.navy.mil -- Dr. Christian A. Hummel NPOI Division, Flagstaff Station US Naval Observatory c/o 3450 Massachusetts Avenue NW Washington DC 20392 Tel.: 202-762 0314 FAX: 202-762 1514 cah@usno.navy.mil